Released on August 18, 2023
- Add
memray
integration (:pr:`8044`) Florian Jetter
- Await
async
listener.stop
inWorker.close
(:pr:`8118`) Hendrik Makait - Minor fixes in
memray
(:pr:`8113`) Florian Jetter - Enable basic
p2p
shuffle fordask-cudf
(:pr:`7743`) Richard (Rick) Zamora - Don't shut down unresponsive workers on
gather()
(:pr:`8101`) crusaderky - Propagate
CancelledError
ingather_from_workers
(:pr:`8089`) crusaderky - Better logging for anomalous task termination (:pr:`8082`) crusaderky
- Handle null partitions in P2P shuffling (:pr:`8116`) Hendrik Makait
- Handle
CancelledError
properly inConnectionPool
(:pr:`8110`) Florian Jetter - Fix additional race condition that can cause P2P restart to deadlock (:pr:`8094`) Hendrik Makait
- Ensure x-axis is uniform when plotting (:pr:`8093`) Florian Jetter
- Fix deadlock in P2P restarts (:pr:`8091`) Hendrik Makait
- Add
memray
integration to API docs (:pr:`8115`) James Bourbeau - Fix default in description of
LocalCluster
sscheduler_port
(:pr:`8073`) Danferno
- Remove
types_mapper
arg now that it is captured infrom_pyarrow_table_dispatch
(:pr:`8114`) Richard (Rick) Zamora - Make P2P shuffle extensible (:pr:`8096`) Hendrik Makait
- Make
PreloadManager
aSequence
(:pr:`8112`) Hendrik Makait - Introduce
PreloadManager
to handle failures in preload setup/teardown (:pr:`8078`) Hendrik Makait - Restructure P2P code (:pr:`8098`) Hendrik Makait
- Make
ToPickle
aGeneric
(:pr:`8097`) Hendrik Makait - Dedicated job for
memray
tests (:pr:`8104`) Florian Jetter - Fix
test_task_groups_update_start_stop
, again (:pr:`8102`) crusaderky - Remove
dumps_task
(:pr:`8067`) Florian Jetter - Simplify usage of queues in nanny (:pr:`6655`) Florian Jetter
- Fix flakiness in tests caused by
WindowsTime
(:pr:`8087`) crusaderky - Overhaul
gather()
(:pr:`7997`) crusaderky - Fix flaky
test_asyncprocess.py::test_simple
(:pr:`8085`) crusaderky - Skip
test_client.py::test_file_descriptors_dont_leak
on Mac OS (:pr:`8080`) Hendrik Makait - Reorder operations in
Worker.close
(:pr:`8076`) Hendrik Makait
Released on August 4, 2023
- Offload CPU intensive sections of update graph to unblock event loop (:pr:`8049`) Florian Jetter
- Log worker close reason in events (:pr:`8042`) Florian Jetter
- Exclude comm handshake from connect timeout (:pr:`7698`) Florian Jetter
- Automatically restart P2P shuffles when output worker leaves (:pr:`7970`) Hendrik Makait
- Add
Client.unregister_scheduler_plugin
method (:pr:`7968`) Brian Phillips - Fix log message (:pr:`8029`) Hendrik Makait
- Send shards grouped by input chunk in P2P rechunking (:pr:`8010`) Hendrik Makait
- Close state machine and add-ins first in
Worker.close
(:pr:`8066`) Hendrik Makait - Fix
decide_worker
picking a closing worker (:pr:`8032`) crusaderky - Raise
CommClosedError
inget_stream_address
(:pr:`8020`) jochenott - Respect average
nthreads
in adaptive (:pr:`8041`) Matthew Rocklin - Use queued tasks in adaptive target (:pr:`8037`) Matthew Rocklin
- Restore support for yield unsafe
Client
context managers and deprecate that support (:pr:`7987`) Thomas Grainger
- Change
worker_saturation
default value to 1.1 in the documention (:pr:`8040`) minhnguyenxuan60 - Clarified
concurrent.futures
section inclient.rst
(:pr:`8048`) mercyo12
- Fix flaky
test_worker_metrics
(:pr:`8069`) crusaderky - Use SPDX in
license
metadata (:pr:`8065`) jakirkham - Rebalance
ci1
markers (:pr:`8061`) Florian Jetter - Ensure stream messages are always ordered (:pr:`8059`) Florian Jetter
- Simplify update graph (:pr:`8047`) Florian Jetter
- Provide close reason when signal is caught (:pr:`8045`) Florian Jetter
- Allow unclosed comms in tests (:pr:`8057`) Florian Jetter
- Cosmetic tweak to
adaptive_target
(:pr:`8052`) crusaderky - Fix linting (:pr:`8046`) Florian Jetter
- Update gpuCI
RAPIDS_VER
to23.10
(:pr:`8033`) - Test against more recent
pyarrow
versions (:pr:`8021`) James Bourbeau - Add a test for
GraphLayout
withscatter
(:pr:`8025`) Irina Truong - Fix compatibility variable naming (:pr:`8030`) Hendrik Makait
Released on July 20, 2023
gather_dep
should handleCancelledError
(:pr:`8013`) crusaderky- Pass
stimulus_id
toSchedulerPlugin.remove_worker
andSchedulerPlugin.transition
(:pr:`7974`) Hendrik Makait - Log
stimulus_id
inretire_worker
(:pr:`8003`) crusaderky - Use
BufferOutputStream
in P2P (:pr:`7991`) Florian Jetter - Add Coiled to ignored modules for code sniffing (:pr:`7986`) Matthew Rocklin
- Progress bar can group tasks by span (:pr:`7952`) Irina Truong
- Improved error messages for P2P shuffling (:pr:`7979`) Hendrik Makait
- Reduce removing comms log to debug level (:pr:`7972`) Florian Jetter
- Fix for
TypeError: '<' not supported
in graph dashboard (:pr:`8017`) Irina Truong - Fix shuffle code to work with
pyarrow
13 (:pr:`8009`) Joris Van den Bossche
- Add some top-level exposition to the p2p rechunking code (:pr:`7978`) Lawrence Mitchell
- Add test when not
repartitioning
forp2p
inset_index
(:pr:`8016`) Patrick Hoefler - Bump
JamesIves/github-pages-deploy-action
from 4.4.2 to 4.4.3 (:pr:`8008`) - Configure asyncio loop using
loop_factory
kwarg rather than using theset_event_loop_policy
(:pr:`7969`) Thomas Grainger - Fix P2P worker cleanup (:pr:`7981`) Hendrik Makait
- Skip
click
v8.1.4 in mypypre-commit
hook (:pr:`7989`) Thomas Grainger - Remove accidental duplicated conversion of
pyarrow
Table
to pandas (:pr:`7983`) Joris Van den Bossche
Released on July 7, 2023
- Propagate spans to tasks (:pr:`7898`) crusaderky
- Make Fine Performance Metrics bar graph horizontal (:pr:`7966`) crusaderky
- Don't pile up
context_meter
callbacks (:pr:`7961`) crusaderky - Polish Fine Performance Metrics plot (:pr:`7963`) crusaderky
- Sign
task-erred
withrun_id
and reject outdated responses (:pr:`7933`) Hendrik Makait - Set
Client.as_current
when entering ctx (:pr:`6527`) Florian Jetter - Re-run erred task on
ComputeTaskEvent
(:pr:`7967`) Hendrik Makait
- Fix crash in spans when
time()
is not monotonic (:pr:`7960`) crusaderky
- Documentation for Fine Performance Metrics and Spans (:pr:`7945`) crusaderky
- Update
client.py
to be consistent with the docstring (:pr:`7705`) Sultan Orazbayev
- Use
distributed.wait_for
intest_close_async_task_handles_cancellation
(:pr:`7955`) Thomas Grainger - Fix flaky UCX tests (:pr:`7950`) Peter Andreas Entschev
Released on June 26, 2023
- Add idle time to fine performance metrics (:pr:`7938`) crusaderky
- Spans: capture code snippets (:pr:`7930`) crusaderky
- Improve memory footprint of P2P rechunking (:pr:`7897`) Hendrik Makait
- Improve error message on invalid state in
_handle_remove_replicas
(:pr:`7920`) Hendrik Makait - Make
ShuffleSchedulerExtension.remove_worker
more robust (:pr:`7921`) Hendrik Makait - Provide more information if occupancy drops below zero (:pr:`7924`) Hendrik Makait
- Improved conversion between
pyarrow
andpandas
in P2P shuffling (:pr:`7896`) Hendrik Makait
- Add
Cluster.called_from_running_loop
and fixCluster.asynchronous
(:pr:`7941`) Jacob Tomlinson - Fix annotations and spans leaking between threads (:pr:`7935`) Irina Truong
- Handle null partitions in P2P shuffling (:pr:`7922`) Jonathan De Troye
- Fix race condition in Fine Performance Metrics sync (:pr:`7927`) crusaderky
- Avoid (:pr:`7923`) by starting
run_id
at 1 (:pr:`7925`) Hendrik Makait - Fix glitches in Fine Performance Metrics stacked graph (:pr:`7919`) crusaderky
- Wipe the cache after (:pr:`7935`) (:pr:`7946`) crusaderky
- Remove grace period for unclosed comms in
gen_cluster
(:pr:`7937`) Thomas Grainger raise pytest.skip
is redundant (:pr:`7939`) crusaderky- Fix
test_rechunk_with_{fully|partially}_unknown_dimension
on CI (:pr:`7934`) Hendrik Makait - Fix compatibility with
numpy
1.25 (:pr:`7932`) crusaderky - Spans: refactor sums of mappings (:pr:`7918`) crusaderky
- Fix flaky
test_send_metrics_to_scheduler
(:pr:`7931`) crusaderky - Avoid calls to
make_current()
andmake_clear()
by usingasyncio.run
inLoopRunner
(:pr:`7467`) Thomas Grainger - Add
needs triage
label to re/opened PRs and issues (:pr:`7916`) Miles - Remove
span_id
from global metrics on scheduler (:pr:`7917`) crusaderky - Add spans to Fine Performance Metrics bokeh dashboard (:pr:`7911`) crusaderky
- FinePerformanceMetrics dashboard overhaul (:pr:`7910`) crusaderky
- Check for
skip-caching
label (:pr:`7907`) Miles - Fix CI changes from (:pr:`7902`) (:pr:`7905`) Hendrik Makait
- Rename
get_default_shuffle_algorithm
toget_default_shuffle_method
(:pr:`7902`) Hendrik Makait - Bump actions/checkout from 3.5.2 to 3.5.3 (:pr:`7904`)
- Refactor P2P rechunk validation (:pr:`7890`) Hendrik Makait
Released on June 9, 2023
- Post fine performance metrics to spans (:pr:`7885`) crusaderky
- Unique Spans (:pr:`7882`) crusaderky
- Add a
timeout
toclient.as_completed
that mirrorsconcurrent.futures.as_completed
timeout
(:pr:`7811`) Thomas Grainger - Enforce dtypes in P2P shuffle (:pr:`7879`) Hendrik Makait
- Support
load=
keyword forClient.upload_file
(:pr:`7873`) James Bourbeau - Support
get_worker()
andworker_client()
in async tasks (:pr:`7844`) Thomas Grainger - Capture line number for code frames (:pr:`7786`) Miles
- Avoid meta roundtrip in P2P shuffle (:pr:`7895`) Hendrik Makait
- Fix Fine Performance Metrics mis-aligned
ColumnData
lengths (:pr:`7893`) Miles - Fix Fine Performance Metrics spilling crash (:pr:`7878`) Miles
- Fix spans bug when
scatter
orclient_desires_new_key
creates a task (:pr:`7886`) crusaderky - Fix Fine Performance Metrics w/ Bokeh 3 (:pr:`7874`) Miles
TaskGroup.start
can move backwards (:pr:`7867`) crusaderky- Use properly imported
MatDescriptor
forcupy
dispatch registration (:pr:`7868`) Charles Blackmon-Luca - Ensure
retire_workers
works if AMM extension hasn't been loaded (:pr:`7863`) crusaderky
- Review user-defined fine performance metrics (:pr:`7894`) crusaderky
- Fix tests that disable the shuffle extension (:pr:`7883`) crusaderky
- Refactor
Scheduler.is_idle
(:pr:`7881`) crusaderky - Link TaskGroups to Spans (:pr:`7869`) crusaderky
- Spans skeleton (:pr:`7862`) crusaderky
- Update gpuCI
RAPIDS_VER
to23.08
(:pr:`7855`) - Bump
JamesIves/github-pages-deploy-action
from 4.4.1 to 4.4.2 (:pr:`7865`)
Released on May 26, 2023
Note
This release drops support for Python 3.8. As of this release Dask supports Python 3.9, 3.10, and 3.11. See this community issue for more details.
- Exclude IPython code from computations (:pr:`7788`) Miles
- Drop Python 3.8 support (:pr:`7840`) Thomas Grainger
- Add
storage_options
toperformance_report
(:pr:`7636`) ypogorelova - Don't warn about mismatched
msgpack
(:pr:`7839`) Irina Truong - Clean up
sys.path
onServer
shutdown (:pr:`7838`) James Bourbeau - Dashboard: Fine Performance Metrics (:pr:`7725`) Miles
- Properly handle unknown chunk sizes in P2P rechunking (:pr:`7856`) Hendrik Makait
- Minimal change to work around (:issue:`7726`) / support for UCX (:pr:`7851`) Benjamin Zaitlen
- Don't end computations until cluster is truly idle (:pr:`7790`) crusaderky
- Explicitly install
anaconda-client
from conda-forge when uploading conda nightlies (:pr:`7861`) Charles Blackmon-Luca - Fix
is_idle
docs build (:pr:`7854`) James Bourbeau - Add tests for P2P barrier fusion (:pr:`7845`) Hendrik Makait
- Avoid
DeprecationWarning
incupy
dispatch registration (:pr:`7836`) Lawrence Mitchell
Released on May 12, 2023
Client.upload_file
send to both Workers and Scheduler and rename scratch directory (:pr:`7802`) Miles- Allow dashboard to be used with bokeh prereleases (:pr:`7814`) James Bourbeau
- Ensure log_event of non-msgpack serializable object do not kill servers (:pr:`7472`) Florian Jetter
- Fix
test_nanny.py
duplicatedpytestmark
definitions (:pr:`7819`) Thomas Grainger - Fix flaky
test_dask_worker.py::test_single_executable_deprecated
(:pr:`7817`) Thomas Grainger
- Annotation-less P2P shuffling (:pr:`7801`) Hendrik Makait
- Fix docstring for
batch_size
inclient.map
(:pr:`7833`) David Chudzicki - Refactor
test_protocol.py
(:pr:`7829`) crusaderky - Lint #6496 (:pr:`7828`) crusaderky
- Remove hardcoded 60s timeout (:pr:`6496`) Florian Jetter
- Add
__init__.py
files to template and static directories (:pr:`7809`) Thomas Grainger - Disable compression for fast comms (:pr:`7768`) crusaderky
- Avoid deprecated
pd.api.types.is_sparse
(:pr:`7813`) James Bourbeau - Bump gpuCI
PYTHON_VER
from 3.8 to 3.9 (:pr:`7812`) Charles Blackmon-Luca
Released on April 28, 2023
- Enable GIL monitoring when gilknocker installed (:pr:`7730`) Miles
- By default only set logging handler if no other handler has been set to avoid double logging (:pr:`7750`) Thomas Grainger
- Cluster wait (:pr:`6700`) Iain Dorrington
- Add Prometheus counter for
SystemMonitor.last_time
(:pr:`7785`) Miles
- Partial revert defaultclient config setting (:pr:`7803`) Florian Jetter
- Delay awaiting async
SchedulerPlugin.{add|remove}_worker
hooks in order to immediately execute all sync ones (:pr:`7799`) Hendrik Makait - Fix
check_idle
not returning the correct value if no change to idleness (:pr:`7781`) Jacob Tomlinson
- Avoid warning when
gilknocker
not installed (:pr:`7808`) James Bourbeau - Only set worker/nanny to
Status.running
if it is inStatus.init
(:pr:`7773`) Thomas Grainger - Add
--cov-config=pyproject.toml
so config is always correctly loaded bypytest-cov
(:pr:`7793`) Thomas Grainger gilknocker
from conda-forge (:pr:`7791`) James Bourbeau- Minor
zict
cleanup (:pr:`7783`) crusaderky - Bump
actions/checkout
from 3.5.0 to 3.5.2 (:pr:`7784`) - Fix typing now that code is tuple of frame(s) (:pr:`7778`) Nat Tabris
Released on April 14, 2023
Note
With this release we are making a change which will require the Dask scheduler to have consistent software and hardware capabilities as the client and workers.
It's always been recommended that your client and workers have a consistent software and hardware environment so that data structures and dependencies can be pickled and passed between them. However recent changes to the Dask scheduler mean that we now also require your scheduler to have the same consistent environment as everything else.
- Meter queue time to the offload executor (:pr:`7758`) crusaderky
- Add GIL contention metric to Prometheus (:pr:`7651`) Miles
- Add methods
Client.forward_logging()
andClient.unforward_logging()
. (:pr:`7276`) Max Bane - Optionally capture more frames in computations (:pr:`7656`) Gabe Joseph
- Consider Jupyter activity in idle timeout (:pr:`7687`) Gabe Joseph
- Add a dashboard component that displays RMM memory (:pr:`7718`) Peter Andreas Entschev
- Improve error message if
shuffle
/rechunk
lost annotations (:pr:`7707`) Hendrik Makait - Exception chaining in P2P shuffling (:pr:`7706`) Hendrik Makait
- Use pickle for graph submissions from client to scheduler (:pr:`7564`) Florian Jetter
- Fix crash on missing env var in dashboard link formatting (:pr:`7729`) Miles
- Fix
randbytes()
on Python 3.8 (:pr:`7771`) crusaderky - Run scheduler of
SubprocessCluster
in subprocess (:pr:`7727`) Hendrik Makait - Drop id from RMM dashboard component (:pr:`7739`) James Bourbeau
- Bump
peter-evans/create-pull-request
from 4 to 5 (:pr:`7766`) - Fix flaky
test_malloc_trim_threshold
in CI (:pr:`7764`) crusaderky - Minor polish in
spill
andworker_memory_manager
(:pr:`7752`) crusaderky - Merge identical
tool.mypy.overrides
sections (:pr:`7749`) Thomas Grainger - Add changelog section for 2023.3.2.1 (:pr:`7755`) Charles Blackmon-Luca
- Specify
ts
resolution explicitly intest_processing_chain
(:pr:`7744`) Patrick Hoefler - Unignore Sphinx
ref.python
(:pr:`7713`) Thomas Grainger - Temporary fix for
test_merge_by_multiple_columns
with pandas 2.0 (:pr:`7747`) James Bourbeau - Remove
dask/gpu
from gpuCI update reviewers (:pr:`7741`) Charles Blackmon-Luca - Update gpuCI
RAPIDS_VER
to23.06
(:pr:`7728`) - Remove test for
DataFrame.to_hdf
(:pr:`7735`) Hendrik Makait - Test P2P shuffling with
DataFrame.to_hdf
(:pr:`7720`) Hendrik Makait scheduler.py
typing - removeallow_incomplete_defs
(:pr:`7721`) Florian Jetter- Remove
bokeh
upper bound (:pr:`7413`) James Bourbeau - Use declarative
setuptools
(:pr:`7629`) Thomas Grainger - Store performance metrics on scheduler (:pr:`7701`) Miles
- Upgrade readthedocs config to ubuntu 22.04 and Python 3.11 (:pr:`7722`) Thomas Grainger
- Clean up legacy cruft from worker reconnection (:pr:`7712`) crusaderky
- Bump
actions/checkout
from 3.4.0 to 3.5.0 (:pr:`7711`) - Drop support for zict 2.1.0 (:pr:`7709`) crusaderky
- Fix
mypy
warning intest_client.py
(:pr:`7710`) crusaderky - Test P2P shuffling with
DataFrame.categorize
(:pr:`7708`) Hendrik Makait
Released on April 5, 2023
- Register atexit handler before Distributed handlers to unblock hanging UCX clusters Lawrence Mitchell Ben Zaitlen
Released on March 24, 2023
- Enhanced thread-safety in
zict.File
(:pr:`7691`) crusaderky - Future deserialization without available client (:pr:`7580`) Florian Jetter
- Support adjusting GIL monitoring interval (:pr:`7650`) Miles
- Gracefully stop GIL monitoring if running (:pr:`7652`) Miles
- Fine performance metrics for
execute
,gather_dep
, etc. (:pr:`7586`) crusaderky - Add GIL metric to dashboard (:pr:`7646`) Miles
- Expose scheduler idle via RPC and HTTP API (:pr:`7642`) Jacob Tomlinson
- Add full dashboard link in scheduler logs (:pr:`7631`) Miles
- Tell workers when their peers have left (so they don't hang fetching data from them) (:pr:`7574`) Thomas Grainger
- Fix regression in dashboard after (:pr:`7586`) (:pr:`7683`) crusaderky
- Fix
OverflowError
inCluster._sync_cluster_info()
(:pr:`7648`) Hendrik Makait - Ensure that serialized data is measured correctly (:pr:`7593`) Florian Jetter
- Fix unexpected indentation in
Client.cancel
docstring (:pr:`7694`) Thomas Grainger - Improve plugin API documentation (:pr:`7653`) Florian Jetter
- Configure sphinx warnings as errors (:pr:`7697`) Thomas Grainger
- Fix naming comparison in
test-report
workflow script (:pr:`7695`) Miles - Temporarily restrict
ipywidgets<8.0.5
(:pr:`7693`) crusaderky - Bump
actions/checkout
from 3.3.0 to 3.4.0 (:pr:`7685`) - Temporarily restrict
ipykernel<6.22.0
(:pr:`7689`) James Bourbeau - Fix typo in
CODEOWNERS
(:pr:`7670`) Hendrik Makait - Avoid
bool
object has no attributeclose
in@gen_cluster
(:pr:`7657`) Thomas Grainger - Fix failing
test_server_close_stops_gil_monitoring
(:pr:`7659`) James Bourbeau - Add
CODEOWNERS
file (:pr:`7645`) Jacob Tomlinson - Remove
weakref
finalizer for Offload Executor (:pr:`7644`) Florian Jetter
Released on March 10, 2023
- Add Jupyter link to dashboard menu if
--jupyter
flag is set (:pr:`7638`) Jacob Tomlinson - Bump minimum
click
version from 7.0 to 8.0 (:pr:`7637`) Miles - Extend
dask
metapackage dependencies (:pr:`7630`) James Bourbeau - Further improvements to
Client.restart_workers
(:pr:`7620`) Miles - P2P offload
get_output_partition
(:pr:`7587`) Florian Jetter - Initial integration of GIL contention metric (:pr:`7624`) Miles
- Add dashboard documentation links (:pr:`7610`) Miles
- Rename shuffle/rechunk config option/kwarg to method (:pr:`7623`) Hendrik Makait
- Return results in
restart_workers
(:pr:`7606`) Miles - Ensure client key cancellation uses ordered messages (:pr:`7583`) Florian Jetter
- Fix undefined
async_wait_for
->async_poll_for
(:pr:`7627`) Miles - Don't send client heartbeat without a
scheduler_comm
(:pr:`7612`) James Bourbeau - Do not unspill on free-keys (:pr:`7607`) crusaderky
- Add notes to
Client.submit
,Client.map
, andClient.scatter
with the description of the current task graph resolution algorithm limitations (:pr:`7588`) Eugene Druzhynin
- Use
range
withpickle
protocol
versions (:pr:`7635`) jakirkham - Share thread pool among P2P shuffle runs (:pr:`7621`) Hendrik Makait
- Replace
psutil
suspend withBlockedGatherDep
intest_failing_worker_with_additional_replicas_on_cluster
(:pr:`7633`) Thomas Grainger - Ignore
pkg_resources
DeprecationWarning
for mindeps (:pr:`7626`) Miles - Implement
wait_for
usingasyncio.timeout()
on 3.11 (:pr:`7571`) Thomas Grainger - Use
tmp_path
fixture instead of outdatedtmpdir
fixture (:pr:`7582`) ypogorelova - Only one
crick
callback (:pr:`7614`) crusaderky - Add mindeps +
numpy
job to tests CI (:pr:`7609`) Miles - Do not
xfail
whole tests due to (:pr:`6705`) (:pr:`7611`) crusaderky
Released on March 1, 2023
- Remove
pyarrow
dependency for rechunking (:pr:`7604`) Florian Jetter - Update
rechunk_transfer
andrechunk_unpack
errors (:pr:`7600`) James Bourbeau
- Remove dead code and document arguments to
ShardBuffer
constructors (:pr:`7590`) Lawrence Mitchell - Fix tests for p2p by default (:pr:`7595`) Florian Jetter
- Remove obsolete cast (:pr:`7596`) Florian Jetter
Released on February 24, 2023
- P2P for array rechunking (:pr:`7534`) Hendrik Makait
- P2P HashJoin (:pr:`7514`) Florian Jetter
- Unpickle Events, Variables, Queues and Semaphore safely without Client context (:pr:`7579`) Florian Jetter
- Allow pickle to fall back to dask_serialize (:pr:`7567`) Florian Jetter
- make
ConnectionPool.remove
cancel connection attempts (:pr:`7547`) Thomas Grainger - Meter how long each task prefix stays in each state (:pr:`7560`) crusaderky
- Avoid parsing
sys.argv
when startingjupyter
server (:pr:`7573`) Brett Naul str
/bytes
compatibility for PyNVML device name (:pr:`7563`) James Bourbeaumetrics.monotonic()
is not monotonic on Windows (:pr:`7558`) crusaderky- Fix for
bytes
/str
discrepancy after PyNVML update (:pr:`7544`) Peter Andreas Entschev
- Raise when attempting P2P with active fuse optimization (:pr:`7585`) Hendrik Makait
- Fix
test_shuffling
(:pr:`7581`) Hendrik Makait - P2P: raise RuntimeError if pyarrow version is not sufficient (:pr:`7578`) Florian Jetter
- Check for dtype support in p2p (:pr:`7425`) Hendrik Makait
- Update parsing of FULL_RAPIDS_VER/FULL_UCX_PY_VER (:pr:`7568`) Charles Blackmon-Luca
- move retry from get_data_from_worker to gather_from_workers (:pr:`7546`) Thomas Grainger
- Increase
numpy
andpandas
version pins for nightlies (:pr:`7562`) James Bourbeau - Set validate=True in all tests (:pr:`7557`) crusaderky
- Remove dead code from _get_task_finished_msg (:pr:`7561`) crusaderky
- Mark tests that take >2s as slow (:pr:`7556`) crusaderky
- Fix test_scatter_no_workers on slow CI (:pr:`7559`) crusaderky
- Unskip
test_delete_some_results
(:pr:`7508`) Hendrik Makait - scatter() should not sidestep the worker transition machinery (:pr:`7545`) crusaderky
- pre-commit bump (:pr:`7541`) crusaderky
- Better assertions in Worker.validate_state() (:pr:`7549`) crusaderky
- Bump jacobtomlinson/gha-find-replace from 2 to 3 (:pr:`7540`) James Bourbeau
- Bump
black
to 23.1.0 (:pr:`7542`) crusaderky - Run GPU tests on python 3.8 & 3.10 (:pr:`7537`) Charles Blackmon-Luca
Released on February 10, 2023
- Rate limit the worker memory logs (:pr:`7529`) Florian Jetter
- Move P2P barrier logic to scheduler extension (:pr:`7519`) Hendrik Makait
- Use PEP 673
Self
type (:pr:`7530`) Thomas Grainger - Tentatively fix
test_pause_while_spilling
(:pr:`7517`) crusaderky - Annotate
asyncio_tcp.py
(:pr:`7522`) crusaderky - Use dask git tip for
mypy
(:pr:`7516`) crusaderky - Upgrade to
mypy
v1 (:pr:`7525`) Thomas Grainger - Clean up calls to
captured_logger
(:pr:`7521`) crusaderky - Update
isort
version to 5.12.0 (:pr:`7513`) Lawrence Mitchell
Released on January 27, 2023
- P2P shuffle deduplicates data and can be run several times (:pr:`7486`) Hendrik Makait
- Reverse order of
get_logs()
andget_worker_logs()
(:pr:`7475`) Nicholas R. Knezek - Add prometheus metric for time and memory used per task prefix (:pr:`7406`) Thomas Grainger
- Additive worker counts in Prometheus (:pr:`7468`) crusaderky
- Add help tool for taskstream (:pr:`7478`) Florian Jetter
- Do not allow for a worker to reject a drop replica request (:pr:`7490`) Hendrik Makait
- Fix un/packing for namedtuples with custom constructors (:pr:`7465`) antonymayi
- Remove
timeout=
from docstring example forworker_client
(:pr:`7497`) Florian Jetter
- Ignore get_default_shuffle_algorithm linting issue (:pr:`7506`) Florian Jetter
- Remove set_config when using default client (:pr:`7482`) Florian Jetter
- Update gpuCI
RAPIDS_VER
to23.04
(:pr:`7501`) - Fix
test_balance_expensive_tasks
and improve helper functions intest_steal.py
(:pr:`7253`) Hendrik Makait - Sign every compute task with run ID to correlate response (:pr:`7463`) Hendrik Makait
Released on January 13, 2023
- Add local
SubprocessCluster
that runs workers in separate processes (:pr:`7431`) Hendrik Makait
- Ensure client session is quiet after
cluster.close()
orclient.shutdown()
(:pr:`7429`) James Bourbeau - Set
lifetime-stagger
default value toNone
(:pr:`7445`) bstadlbauer - Memory thresholds should never be exactly
0.0
(:pr:`7458`) Stuart Berg - Remove the Incorrect-Sizeof-Warning (:pr:`7450`) Mads R. B. Kristensen
- Log exceptions in P2P shuffle tasks (:pr:`7442`) Hendrik Makait
- Add support for packing
namedtuple
and add test for future resolution in submit (:pr:`7292`) Andrew - Avoid deep copy on
lz4
decompression (:pr:`7437`) crusaderky - Avoid deep copy of
numpy
buffers on unspill (:pr:`7435`) crusaderky - Don't error when clicking on empty task stream plot (:pr:`7432`) James Bourbeau
- Do not count spilled memory when comparing vs. process memory (:pr:`7430`) crusaderky
- Stop
Client
periodic callbacks duringshutdown()
(:pr:`7428`) James Bourbeau - Add
dask spec
CLI (:pr:`7427`) Matthew Rocklin - Create new
zstd
(de)compressor for each compression call (:pr:`7404`) Dylan Wragge - Rename
managed_in_memory
etc. to match GUI (:pr:`7418`) crusaderky - Warn users when
sizeof()
returns inflated output (:pr:`7419`) crusaderky
- Ensure dicts are properly recognized as
msgpack
serializable (:pr:`7473`) Florian Jetter - Reset state of
ShuffleSchedulerExtension
on restart (:pr:`7446`) Hendrik Makait - Reject non-string column names in P2P shuffle (:pr:`7447`) Hendrik Makait
- Avoid
int32
in dashboard (:pr:`7443`) Matthew Rocklin - Fix
P2PShuffle
serialization for categorical data (:pr:`7410`) Hendrik Makait WorkerPorcess
blocks on kill if still starting (:pr:`7424`) Matthew Rocklin
- Move Prometheus docs from
dask/dask
(:pr:`7405`) crusaderky
- Various cleanups in semaphore (:pr:`5885`) Florian Jetter
test_rlimit
fails on MacOSX (:pr:`7457`) crusaderky- Bump
actions/checkout
from 3.2.0 to 3.3.0 (:pr:`7464`) - Remove conditional imports of
psutil
(:pr:`7462`) crusaderky - Drop support for
zict < 2.1.0
(:pr:`7456`) crusaderky - Fix flaky
test_digests
(:pr:`7454`) crusaderky - Add minimum dependency testing to CI (:pr:`7285`) Charles Blackmon-Luca
- Avoid overflow in
statitics.mean
(:pr:`7426`) Matthew Rocklin - Ignore
numpy
bool8
deprecation (:pr:`7423`) Matthew Rocklin - Add missing skips for pyarrow (:pr:`7416`) Elliott Sales de Andrade
- Be more permissive about expected ciphers in tests (:pr:`7417`) Elliott Sales de Andrade
- Revert "TST: Fetch executables from build root (:pr:`2551`)" (:pr:`7415`) Elliott Sales de Andrade
Released on December 16, 2022
SpillBuffer
metrics (:pr:`7368`) crusaderky- Prometheus: measure how much spilling blocks the event loop (:pr:`7370`) crusaderky
- Add
transfer_outgoing_bytes_total
metric (:pr:`7388`) Gabe Joseph - Fail
P2PShuffle
gracefully upon worker failure (:pr:`7326`) Hendrik Makait
- Select queued tasks in stimuli, not transitions (:pr:`7402`) Gabe Joseph
- Check
ContextVar
indefault_client
(:pr:`7369`) Matthew Rocklin - Fix sending event messages to non-subscribers (:pr:`7014`) Laurence Watts
- Set sizing mode on
Tabs
to avoid layout collapse (:pr:`7365`) Mateusz Paprocki
- Restructure
P2PShuffle
extensions (:pr:`7390`) Hendrik Makait - Add Python 3.11 classifier (:pr:`7408`) James Bourbeau
- Add support for Python 3.11 (:pr:`7249`) Thomas Grainger
- Add test for using annotations with
client.submit
andclient.map
(:pr:`7399`) James Bourbeau - Bump
actions/checkout
from 3.1.0 to 3.2.0 (:pr:`7393`) - Remove superfluous
ShuffleSchedulerExtension.barriers
(:pr:`7389`) Hendrik Makait - Remove ignore annotation-unchecked (:pr:`7379`) crusaderky
- Remove
tornado
max version from nightly recipe (:pr:`7376`) Charles Blackmon-Luca - Remove the experimental feature warning for
Semaphore
(:pr:`7373`) Florian Jetter
Released on December 2, 2022
- Expose event loop health metrics in Prometheus (:pr:`7360`) Hendrik Makait
- Allow log propagation by default (:pr:`5669`) Florian Jetter
- Clean up of
unpack_remotedata()
(:pr:`7322`) Mads R. B. Kristensen - Upgrade to
tornado
6.2 (:pr:`7286`) Thomas Grainger - Introduce
Server
levelcomm
counters (:pr:`7332`) Florian Jetter - Prometheus debug log (:pr:`7302`) Florian Jetter
- Catch
BaseException
s from user tasks (:pr:`5997`) Gabe Joseph - Impossible use case of erred deps in transition to waiting (:pr:`7354`) crusaderky
- Fix a deadlock when queued tasks are resubmitted quickly in succession (:pr:`7348`) Florian Jetter
- Editorial changes to Prometheus documentation (:pr:`7350`) Hendrik Makait
- Fetch all artifacts (:pr:`7355`) Enrico Minack
- Delay
fsspec
andurllib3
import time (:pr:`6659`) Florian Jetter - Bump
mypy
(:pr:`7349`) crusaderky - Bump
mypy
and remove win specific run (:pr:`7344`) Florian Jetter - Finish overhaul of
SchedulerState
annotations (:pr:`7333`) crusaderky - Fix flaky
test_pause_while_spilling
(:pr:`7334`) Gabe Joseph - Update gpuCI
RAPIDS_VER
to23.02
(:pr:`7337`)
Released on November 18, 2022
- Restrict
bokeh=3
support (:pr:`7329`) Gabe Joseph - Respect death timeout when waiting for scheduler file (:pr:`7296`) Florian Jetter
- Always raise exception if
P2PShuffle
s send fails (:pr:`7317`) Hendrik Makait
- Fix typo in
client.run()
docstring (:pr:`7315`) Richard Pelgrim - Note queuing default change in changelog (:pr:`7314`) Gabe Joseph
- Update
ga-yaml-parser
step in gpuCI updating workflow (:pr:`7335`) Charles Blackmon-Luca - Remove exception handling from transitions (:pr:`7316`) crusaderky
- Turn private functions into private
SchedulerState
methods (:pr:`7260`) Hendrik Makait - Bump
toolz
minimum version to0.10.0
(:pr:`7309`) Sam Grayson
Released on November 15, 2022
Note
This release changes the default scheduling mode to use :ref:`queuing <queuing>`. This will significantly reduce cluster memory use in most cases, and generally improve stability and performance. Learn more here and please provide any feedback on this discussion.
In rare cases, this could make some workloads slower. See the :ref:`documentation <adjust-queuing>` for more information, and how to switch back to the old mode.
- Add
ForwardOutput
worker plugin to forwardstdout
andstderr
to client. (:pr:`7297`) Hendrik Makait - Duration counters on prefix level (:pr:`7288`) Florian Jetter
- Include button for launching JupyterLab layout in repr (:pr:`7218`) Ian Rose
- Support MIG parsing during CUDA context creation in UCX initialization (:pr:`6720`) Peter Andreas Entschev
- Handle
/metrics
endpoint withoutprometheus-client
installed (:pr:`7234`) Hendrik Makait - Add support for unpacking namedtuples in remote data (:pr:`7282`) Andrew
- Enable queuing by default (:pr:`7279`) Florian Jetter
- Fix
exists
->``exist`` typo in scheduler error messages (:pr:`7281`) Matthew Plough - If there's an exception in the
Client
async context manager body then close fast (:pr:`6920`) Thomas Grainger
- Copyediting typos +
codespell
pre-commit
hook for docs (:pr:`7294`) Matthew Plough - Queuing docs (:pr:`7203`) Gabe Joseph
- Ensure category is optional when logging
"warn"
events (:pr:`7169`) James Bourbeau - Edge and impossible transitions to memory (:pr:`7205`) crusaderky
- Use
conda-incubator/setup-miniconda@v2.2.0
(:pr:`7310`) jakirkham - Allow
bokeh=3
(:pr:`5648`) James Bourbeau - Fix typos in P2P shuffle code (:pr:`7304`) Hendrik Makait
- Reenable
test_bad_disk
(:pr:`7300`) Florian Jetter - Reduce max-runs in test reports (:pr:`7299`) Florian Jetter
- Revert idle classification when
worker-saturation
is set (:pr:`7278`) Florian Jetter - Fix flaky
deadline_expiration
(:pr:`7287`) Florian Jetter - Rewrite of P2P control flow (:pr:`7268`) Florian Jetter
- Add codecov token (:pr:`7277`) Florian Jetter
- Bump minimum
bokeh
version to 2.4.2 (:pr:`7271`) James Bourbeau - Remove deprecated code calls to
IOLoop.make_current()
(:pr:`7240`) Thomas Grainger - Improved test for balancing expensive tasks (:pr:`7272`) Hendrik Makait
- Refactor
semaphore._Watch
into general-purposeDeadline
utility (:pr:`7238`) Hendrik Makait - Coverage report fixing (:pr:`7270`) Tom Hu
- Require Click 7.0+ (:pr:`7226`) jakirkham
- Drop tests (:pr:`7269`) Hendrik Makait
- Replace
test_(do_not_)steal_communication_heavy_tasks
tests with more robust versions (:pr:`7243`) Hendrik Makait xfail
test_bad_disk
(:pr:`7265`) crusaderky- Move
transition_log
fromScheduler
toSchedulerState
(:pr:`7254`) crusaderky - Remove
Scheduler.log
(:pr:`7258`) crusaderky - Use latest
pickle
(:pr:`5826`) jakirkham - Polish parsing of
worker-saturation
from config (:pr:`7255`) crusaderky - Avoid expensive occupancy calculation when unused (:pr:`7257`) Gabe Joseph
- Un-skip
test_nested_compute
(:pr:`7247`) Gabe Joseph - Review
test_do_not_steal_communication_heavy_tasks
(:pr:`7250`) crusaderky - Fix
test_stress_creation_and_deletion
(:pr:`7215`) crusaderky - Raise exceptions in
Server.handle_stream
instead of swallowing/logging (:pr:`7162`) Hendrik Makait - Upgrade to
mypy
v0.982 (:pr:`7241`) Thomas Grainger - Fix
_update_scheduler_info
hanging failed tests (:pr:`7225`) Gabe Joseph - Bump
xarray-contrib/ci-trigger
from 1.1 to 1.2 (:pr:`7232`)
Released on October 31, 2022
- Reverted a bug where Bokeh was accidentally made non-optional (:pr:`7230`) Oliver Holworthy
- Schedule a queued task when a task secedes (:pr:`7224`) Gabe Joseph
This was a hotfix release
Released on October 28, 2022
- Add
Client.restart_workers
method (:pr:`7154`) James Bourbeau - Implement
PackageInstall
plugin forpip
andconda
(:pr:`7126`) Hendrik Makait
- Add prometheus collector for work-stealing (:pr:`7206`) Hendrik Makait
- Track reason of workers closing and restarting (:pr:`7166`) Hendrik Makait
- Show no-worker on task progress bar (:pr:`7171`) Florian Jetter
- Set
OPENBLAS_NUM_THREADS
by default (:pr:`7177`) James Bourbeau - Optionally provide local directory to data constructor (:pr:`7153`) Lawrence Mitchell
- Introduce
distributed.comm.ucx.environment
config slot (:pr:`7164`) Lawrence Mitchell - Log information about memory limit (:pr:`7160`) Florian Jetter
- Improve log messages on scheduler for restart (:pr:`7150`) Florian Jetter
- More comprehensive
WorkerState
task counters (:pr:`7167`) crusaderky
- Add note to changelog about new CLI (:pr:`7178`) James Bourbeau
- Update AMM docs (:pr:`7158`) Benjamin Zaitlen
- Add
CondaInstall
to plugins doc (:pr:`7149`) James Bourbeau
- Update minimum
bokeh
version message (:pr:`7172`) James Bourbeau - Revamped implementations of remote
print()
andwarn()
, fixing #7095 (:pr:`7129`) Max Bane
- Temporarily restrict
bokeh<3
(:pr:`7219`) James Bourbeau - Make
Scheduler.reschedule
private (:pr:`7216`) crusaderky - Fix
decide_worker_rootish_queuing_disabled
assert (:pr:`7065`) Gabe Joseph - Fix flaky
test_include_communication_in_occupancy
(:pr:`7212`) Gabe Joseph - Do not raise on leaked websockets (:pr:`7199`) Florian Jetter
- Update nightly recipes with CLI tests, dependency changes (:pr:`7201`) Charles Blackmon-Luca
- Make
p2p
shuffle submodules private (:pr:`7186`) Florian Jetter - Backport tornado
PeriodicCallback
(:pr:`7165`) Florian Jetter - Fix
mypy
failure on CI (:pr:`7198`) Florian Jetter - User a layer for
p2p
shuffle (:pr:`7180`) Florian Jetter - Type annotations for shuffle (:pr:`7185`) Florian Jetter
- Do not close worker on comm error in heartbeat (:pr:`7163`) Hendrik Makait
- Errors when setting TCP timeouts log as error (:pr:`7161`) Florian Jetter
- Remove incorrect advice from
pre-commit
config (:pr:`7159`) crusaderky - Bump
the-coding-turtle/ga-yaml-parser
from 0.1.1 to 0.1.2 (:pr:`7146`) - Bump
JamesIves/github-pages-deploy-action
from 4.1.7 to 4.4.1 (:pr:`7145`) - Use functionalities network for codecov uploader (:pr:`7148`) Florian Jetter
- Use counter metric type where appropriate,
incoming_count
was reporting bytes (:pr:`7125`) Nat Tabris
Released on October 14, 2022
Note
This release deprecates dask-scheduler
, dask-worker
, and dask-ssh
CLIs in favor of dask scheduler
, dask worker
, and dask ssh
,
respectively. The old-style CLIs will continue to work for a time, but will be
removed in a future release.
As part of this migration the --reconnect
, --nprocs
, --bokeh
,
--bokeh-port
CLI options have also been removed for both the old- and new-style
CLIs. These options had already previously been deprecated.
- Use of new dask CLI (:pr:`6735`) Doug Davis
- Refactor occupancy (:pr:`7075`) Hendrik Makait
- Expose managed/unmanaged/spilled memory to Prometheus (:pr:`7112`) crusaderky
- Round up
saturation-factor
(:pr:`7116`) Gabe Joseph - Return default on
KeyError
at any level inget_metadata
(:pr:`7109`) Hendrik Makait - Count task states per task prefix and expose to Prometheus (:pr:`7088`) Nat Tabris
- Add
scheduler-sni
option for dask workers (:pr:`6290`) Burt Holzman
- Improve exception catching in UCX communication (:pr:`7132`) Peter Andreas Entschev
- Improve robustness of
PipInstall
plugin (:pr:`7111`) Hendrik Makait
- Fix dependencies that should point to
dask/dask
(:pr:`7138`) James Bourbeau - Hold on to
z.sum()
until test completes (:pr:`7136`) Lawrence Mitchell - Bump
peter-evans/create-pull-request
from 3 to 4 (:pr:`7120`) - Update typing for
system_monitor
afterpython/typeshed#8829
(:pr:`7131`) Lawrence Mitchell - Fix two potentially flaky queuing tests (:pr:`7124`) Gabe Joseph
- Bump
EnricoMi/publish-unit-test-result-action
from 1 to 2 (:pr:`7121`) - Bump
actions/checkout
from 2 to 3.1.0 (:pr:`7119`) - Revamp
SystemMonitor
(:pr:`7097`) crusaderky - Bump
actions/cache
from 2 to 3 (:pr:`7118`) - Bump
actions/upload-artifact
from 2 to 3 (:pr:`7117`) - Move dependabot configuration file (:pr:`7115`) James Bourbeau
- Enable dependabot for GitHub Actions (:pr:`7101`) Florian Jetter
- Update coverage upload action (:pr:`7100`) Florian Jetter
- Adjust hardware benchmarks bokeh test (:pr:`7096`) Florian Jetter
- Multi-platform mypy checks (:pr:`7094`) crusaderky
- Update gpuCI
RAPIDS_VER
to22.12
(:pr:`7084`)
Released on September 30, 2022
- Smarter stealing with dependencies (:pr:`7024`) Hendrik Makait
- Enable Active Memory Manager by default (:pr:`7042`) crusaderky
- Allow timeout strings in
distributed.wait
(:pr:`7081`) James Bourbeau - Make AMM memory measure configurable (:pr:`7062`) crusaderky
- AMM support for actors (:pr:`7072`) crusaderky
- Expose
message-bytes-limit
in config (:pr:`7074`) Hendrik Makait - Detect mismatching Python version in scheduler (:pr:`7018`) Hendrik Makait
- Improve
KilledWorker
message users see (:pr:`7043`) James Bourbeau - Support for cgroups v2 and respect soft limits (:pr:`7051`) Samantha Hughes
- Catch
BaseException
on UCX read error (:pr:`6996`) Peter Andreas Entschev - Fix transfer limiting in
_select_keys_for_gather
(:pr:`7071`) Hendrik Makait - Parse
worker-saturation
if a string (:pr:`7064`) Gabe Joseph Nanny(config=...)
parameter overlays global dask config (:pr:`7069`) crusaderky- Ensure default clients don't propagate to subprocesses (:pr:`7028`) Florian Jetter
- Improve documentation of
message-bytes-limit
(:pr:`7077`) Hendrik Makait - Minor tweaks to Sphinx documentation (:pr:`7041`) crusaderky
- Improve
upload_file
API documentation (:pr:`7040`) Florian Jetter
test_serialize_numba
: Workaround issue withnp.empty_like
in NP 1.23 (:pr:`7089`) Graham Markall- Type platform constants for
mypy
(:pr:`7091`) jakirkham dask-worker-space
(:pr:`7054`) crusaderky- Remove failing test case (:pr:`7087`) Hendrik Makait
test_default_client
(:pr:`7058`) crusaderky- Fix
pre-commit
fails with recent versions ofmypy
andpandas
(:pr:`7068`) crusaderky - Add factorization utility (:pr:`7048`) James Bourbeau
Released on September 16, 2022
- Add dashboard component for size of open data transfers (:pr:`6982`) Hendrik Makait
- Allow very fast keys and very expensive transfers as stealing candidates (:pr:`7022`) Florian Jetter
- No longer double count transfer cost in stealing (:pr:`7036`) Hendrik Makait
- Make
test_wait_first_completed
robust (:pr:`7039`) Florian Jetter - Partial annotations for
SchedulerState
(:pr:`7023`) crusaderky - Add more type annotations to
stealing.py
(:pr:`7009`) Florian Jetter - Update codecov settings (:pr:`7015`) Florian Jetter
- Speed up
test_balance
(:pr:`7008`) Florian Jetter - Fix test report after queuing job added (:pr:`7012`) Gabe Joseph
- Clean up env variables in Gihub Actions (:pr:`7001`) crusaderky
- Make
test_steal_reschedule_reset_in_flight_occupancy
non timing dependent (:pr:`7010`) Florian Jetter - Replaced
distributed.utils.key_split
withdask.utils.key_split
(:pr:`7005`) Luke Conibear - Revert "Revert "Limit incoming data transfers by amount of data" (:pr:`6994)" (:pr:`7007`) Florian Jetter
- CI job running tests with queuing on (:pr:`6989`) Gabe Joseph
- Fix
distributed/tests/test_client_executor.py::test_wait
(:pr:`6990`) Florian Jetter
Released on September 2, 2022
- Limit incoming data transfers by amount of data (:pr:`6975`) Hendrik Makait
- Expose transfer-related metrics in
Worker.get_metrics
andWorkerMetricCollector
(:pr:`6936`) Hendrik Makait - Withhold root tasks (no co assignment) (:pr:`6614`) Gabe Joseph
- Improve differentiation between incoming/outgoing connections and transfers (:pr:`6933`) Hendrik Makait
- Change memory bars color on spilling/paused status (:pr:`6959`) crusaderky
- Ensure restart clears taskgroups et al (:pr:`6944`) Florian Jetter
- Optimise
scheduler.get_comm_cost
set difference (:pr:`6931`) Lawrence Mitchell - Expose setting multiple protocols and ports via the
dask-scheduler
CLI (:pr:`6898`) Jacob Tomlinson - Make
TextProgressBar
clear the line when finished (:pr:`5968`) Vincenzo Eduardo Padulano
- Revert
getaddrinfo
fast path for Python 3.8 (:pr:`6978`) Florian Jetter - cancelled/resumed->long-running transitions (:pr:`6916`) crusaderky
- Deprecate default value for
Client.wait_for_workers
(:pr:`6942`) Florian Jetter
- Document
Scheduler
andWorker
state machine (:pr:`6948`) crusaderky - Insert
memory_limit
parameter intoLocalCluster
docstring (:pr:`6839`) Crislana Rafael
- Revert "Limit incoming data transfers by amount of data" (:pr:`6994`) Florian Jetter
- Cache conda environment between CI test runs (:pr:`6855`) Charles Blackmon-Luca
- Revert "Fix co-assignment for binary operations" (:pr:`6985`) Gabe Joseph
- Cache
test_report
shelves in CI (:pr:`6937`) Florian Jetter - Cleanup
ipywidgets
mocking (:pr:`6918`) Thomas Grainger - Improve testing of
{Scheduler|Worker}MetricCollector
(:pr:`6945`) Hendrik Makait - Clean up nanny
WorkerProcess.kill
(:pr:`6972`) Gabe Joseph - Rewrite
test_reconnect
to use subprocess to kill scheduler reliably (:pr:`6967`) Florian Jetter - Task state domain on the scheduler side (:pr:`6929`) crusaderky
- Remove
@avoid_ci
fromtest_steal
(:pr:`6872`) crusaderky - Use
async with Worker
in tests (:pr:`6958`) crusaderky - Ignore spurious warnings in
test_quiet_close_process
(:pr:`6955`) crusaderky - Fix tests on Windows (:pr:`6954`) Hendrik Makait
- Prevent duplicates in
HeapSet.sorted()
(:pr:`6952`) crusaderky - Propagate worker address and improve
_remove_from_processing
behavior (:pr:`6946`) Hendrik Makait - Add
HeapSet._sorted
internal flag (:pr:`6949`) Gabe Joseph - Add
HeapSet.peekn
(:pr:`6947`) Gabe Joseph - Fix
pyright
error when importing fromdistributed
(:pr:`6904`) Ian Liu Rodrigues - Always return
ws.address
from_remove_from_processing
(:pr:`6884`) Hendrik Makait - Use
async with Client:
in tests (:pr:`6921`) Thomas Grainger - Ensure relative memory limits work as percentage of system memory (:pr:`6923`) Florian Jetter
Released on August 19, 2022
- Drop comparison of versions against all clients (:pr:`6861`) Hendrik Makait
- Log the worker name if set (:pr:`6866`) Johannes Lange
- Skip
getaddrinfo
thread if host is already resolved, usingsocket.AI_NUMERIC*
(:pr:`6847`) Thomas Grainger - Display unexpected state in
Worker.execute
validation (:pr:`6856`) James Bourbeau pre-spawn-environ
(:pr:`6841`) crusaderky- Dump
has_what
,missing_dep_flight
(:pr:`6830`) Gabe Joseph
cancelled
/resumed
->rescheduled
transition (:pr:`6913`) crusaderky- Fix resource deallocation for resumed tasks (:pr:`6914`) crusaderky
- Only close scheduler in
SpecCluster
if it exists (:pr:`6888`) Matthew Rocklin - Fix issue if
exc.reason
isNone
(:pr:`6881`) Hendrik Makait - Always close
BatchedSend
write coroutines (:pr:`6865`) Gabe Joseph - Harden preamble of
Worker.execute
against race conditions (:pr:`6878`) crusaderky
- Fix typo (:pr:`6870`) Pieter Gijsbers
- Use retries for the test report (:pr:`6926`) Florian Jetter
- Duplicated code:
in_flight_tasks
validation (:pr:`6917`) crusaderky ipywidgets
8 compatibility (:pr:`6912`) James Bourbeau- Overhaul transitions for the
resumed
state (:pr:`6699`) crusaderky - Don't upgrade to
ipywidgets
8 (:pr:`6910`) crusaderky - Clean up
cluster
process reaping (:pr:`6840`) Gabe Joseph - Don't use
bokeh
Figure
in tests (:pr:`6721`) Bryan Van de Ven - Work around incompatibility of crick with setuptools 65 (:pr:`6887`) crusaderky
- Add max version constraint for
dask-core
in nightlies (:pr:`6862`) Charles Blackmon-Luca - Replace
port = random.randint(
withport = d.utils.open_port
(:pr:`6883`) Thomas Grainger - Fix flaky
test_wall_clock
(:pr:`6879`) crusaderky - Add descriptive error message to assert (:pr:`6871`) Hendrik Makait
- Increase timeout in
test_quiet_process
(:pr:`6857`) Florian Jetter - Descriptive title for test report (:pr:`6849`) Hendrik Makait
- Add
flake8-bugbear
as plugin topre-commit
(:pr:`6809`) Hendrik Makait - Remove redundant use of
with clean():
(:pr:`6852`) Thomas Grainger - Show actual Job URL on test report (:pr:`6837`) Florian Jetter
- Update
pre-commit
dependencies (:pr:`6851`) Hendrik Makait - Call exit callback even if
AsyncProcess
is reaped elsewhere (:pr:`6684`) Thomas Grainger - Avoid function calls in argument defaults (:pr:`6812`) Hendrik Makait
- Ignore warning for unclose
SSHCluster
in tests (:pr:`6827`) Florian Jetter
Released on August 5, 2022
- Add Jupyter Server to Dask Scheduler (:pr:`6737`) Matthew Rocklin
- Human-readable formatting for disk I/O and renaming to diff net and disk (:pr:`6835`) Hendrik Makait
- Add
Cluster.get_client()
method (:pr:`6745`) Julia Signell - Start bokeh app to activate bokeh's clean session callbacks (:pr:`6728`) Martí Zamora
- Ensure Nanny doesn't restart workers that fail to start, and joins subprocess (:pr:`6427`) Gabe Joseph
- Don't connect to cluster subprocesses at shutdown (:pr:`6829`) Gabe Joseph
- Fix
restart
wait for workers edge case (:pr:`6823`) Gabe Joseph - Fix spilled size calculation in
Slow
(:pr:`6789`) Hendrik Makait
- Deprecate passing stopped loops to
LoopRunner
(and therefore Client/Cluster) (:pr:`6680`) Thomas Grainger
- Add text to top of API docs to make sure that users are exposed to
LocalCluster
early (:pr:`6793`) Julia Signell - Change title for plugins documentation (:pr:`6733`) Sarah Charlotte Johnson
- Only set 5s connect timeout in
gen_cluster
tests (:pr:`6822`) Gabe Joseph - Fix flaky
test_worker_who_has_clears_after_failed_connection
(:pr:`6832`) Gabe Joseph - Add missing skips for pyarrow (:pr:`6787`) Elliott Sales de Andrade
- Miscellaneous
flake8-bugbear
issues (:pr:`6814`) Hendrik Makait - Assert otherwise pointless comparisons (B015) (:pr:`6811`) Hendrik Makait
- Remove unused functions from
utils_test.py
(:pr:`6807`) Hendrik Makait - Fix Jupyter security note (:pr:`6818`) Jacob Tomlinson
- Improve
check_thread_leak
output (:pr:`6797`) Gabe Joseph - Use contextmanager to ensure clients are closed and do not leak (:pr:`6817`) Hendrik Makait
- Robust thread termination in
test_watch
andtest_watch_requires_lock_to_run
(:pr:`6788`) Hendrik Makait - Avoid unused loop control variable or name them
_
(:pr:`6813`) Hendrik Makait - Replace
assert False
where an exception should always be thrown (:pr:`6815`) Hendrik Makait - Avoid mutable argument defaults in tests (:pr:`6810`) Hendrik Makait
- Avoid mutable argument defaults outside of tests (:pr:`6665`) Hendrik Makait
- Update gpuCI
RAPIDS_VER
to22.10
(:pr:`6798`) - Use same Python for dask worker tests (:pr:`6786`) Elliott Sales de Andrade
Released on July 22, 2022
- Dashboard for failed tasks (:pr:`6595`) Ian Rose
- Wait for workers to return in
Client.restart
(:pr:`6714`) Gabe Joseph - Remove global mutable
Cluster._cluster_info
(:pr:`6487`) Thomas Grainger
- Fix: nvml no early init (:pr:`6678`) Lawrence Mitchell
- Fix bug when restarting client (:pr:`6654`) Iain Dorrington
- Failure to spill breaks available resources (:pr:`6703`) crusaderky
- Fix resource allocation for tasks with dependencies (:pr:`6676`) Hendrik Makait
- Revert "Set
MALLOC_TRIM_THRESHOLD_
before interpreter start" (:pr:`6777`) Gabe Joseph - Fix mypy lint in CI (:pr:`6779`) jakirkham
- Remove
test_restart_fast_sync
,test_fast_kill
(:pr:`6750`) Gabe Joseph - Fix flaky
test_async_task_group_call_later_executes_delayed_task_in_background
(:pr:`6744`) Hendrik Makait - Drop redundant
geninc
(:pr:`6740`) Hendrik Makait - Remove unused
worker_coroutines
(:pr:`6739`) Gabe Joseph - Store ready and constrained tasks in heapsets (:pr:`6711`) crusaderky
- Improve tests for cancelled state (:pr:`6717`) crusaderky
- Future-proof Bokeh value import (:pr:`6707`) Bryan Van de Ven
- Revert temporary stress test (:pr:`6712`) crusaderky
- Validate constrained tasks (:pr:`6698`) crusaderky
- Minor quality-of-life tweaks to cancelled state (:pr:`6701`) crusaderky
- Pickle worker state machine exceptions (:pr:`6702`) crusaderky
- Partial matches for worker state machine instructions (:pr:`6704`) crusaderky
- Automatically mark all WorkerState tests (:pr:`6706`) crusaderky
Released on July 8, 2022
- Use a tempdir path by default instead of cwd for the worker scratch dir (:pr:`6658`) Florian Jetter
- Add
WorkerState.all_running_tasks
(:pr:`6690`) Hendrik Makait Scheduler.reschedule()
works only by accident (:pr:`6339`) crusaderky- Remove spurious
UnpauseEvent
at worker start (:pr:`6652`) crusaderky - Log if closing an executor is not possible in thread (:pr:`6644`) Florian Jetter
- Cloudpickle register by value (:pr:`6466`) Ian Rose
- Adding replicas to a task in fetch now sends it to flight immediately (:pr:`6594`) crusaderky
- Fix dump output of parameter-less events (:pr:`6695`) crusaderky
- Set
MALLOC_TRIM_THRESHOLD_
before interpreter start (:pr:`6681`) crusaderky - Fix deadlocks around rescheduled and resumed states (:pr:`6673`) crusaderky
has_arg
returnsTrue
for keyword-only arguments (:pr:`6648`) Lawrence Mitchell- Transitions caused by worker death use old 'worker-connect'
stimulus_id
(:pr:`6657`) crusaderky - A key is forgotten while
acquire-replicas
is running (:pr:`6638`) crusaderky
- Revisit
WorkerState.long_running
set (:pr:`6697`) crusaderky WorkerState
unit tests for resumed state (:pr:`6688`) crusaderky- Bump version of pandas-stubs (:pr:`6691`) crusaderky
- Add
dummy
factory methods forExecuteSuccessEvent
andExecuteFailureEvent
(:pr:`6687`) Hendrik Makait - Pin
tornado<6.2
in nightly conda recipes (:pr:`6675`) Peter Andreas Entschev - Refactor resource restriction handling in
WorkerState
(:pr:`6672`) Hendrik Makait test_signal
andtest_terminate
occasionally returnSIGKILL
on MacOS (:pr:`6671`) crusaderky- Use the
loop
fixture in even more tests (:pr:`6674`) Thomas Grainger - Unconditionally
import ssl
(:pr:`6670`) Thomas Grainger - Use the
loop
fixture in more tests (:pr:`6642`) Thomas Grainger - Pin tornado to <6.2 (:pr:`6668`) Florian Jetter
- Handle
AsyncTaskGroupClosedError
(:pr:`6664`) Hendrik Makait - Replace occurrences of large delay
slowinc
with locks (:pr:`6656`) Florian Jetter - Merge
extend-ignore
andignore
values forflake8
(:pr:`6660`) Hendrik Makait - Remove server close background task grace period (:pr:`6633`) Thomas Grainger
- Do not use tempfile in
utils_test
(:pr:`6651`) Florian Jetter close_worker
cleanup (:pr:`6650`) crusaderky- Rewrite
test_cancelled_resumed_after_flight_with_dependencies
usingWorkerState
(:pr:`6645`) crusaderky - Log the actual duration to create a directory (:pr:`6647`) Florian Jetter
pandas
type stubs (:pr:`6635`) crusaderky- Remove unused
__started
Event
inServer
(:pr:`6615`) Florian Jetter - Use safe temp directory in
gen_cluster
(:pr:`6628`) Florian Jetter - Print CI host info (:pr:`6629`) crusaderky
- Deduplicate
data_needed
(:pr:`6587`) crusaderky - Remove
EnsureCommunicatingAfterTransitions
(:pr:`6462`) crusaderky - Pickle
WorkerState
(:pr:`6623`) crusaderky - Harden vs.
TaskState
collisions (:pr:`6593`) crusaderky - Do not interact with the event loop when the cluster is garbage collected (:pr:`6627`) Thomas Grainger
Released on June 24, 2022
This release includes the Worker State Machine refactor. The expectation should be that the worker state is its own synchronous subclass. Pulling all the state out into its own class allows us to write targeted unit tests without invoking any concurrent or asynchronous code.
See :pr:`5736` for more information.
- Make worker state machine methods private (:pr:`6564`) crusaderky
- Yank state machine out of Worker class (:pr:`6566`) crusaderky
- Track
worker_state_machine.TaskState
instances (:pr:`6525`) Hendrik Makait - Trivial tweaks to the Worker State Machine (:pr:`6586`) crusaderky
- Replace
loop.call_later
andloop.add_callback
with background tasks added to Server. (:pr:`6603`) Thomas Grainger - Support for neater
WorkerState
tests (:pr:`6609`) crusaderky - Limit TCP writes with Tornado to 2GB (:pr:`6557`) hhuuggoo
- Enable
no_implicit_optional
for scheduler (:pr:`6622`) Thomas Grainger
- Partial revert of compute-task message format (:pr:`6626`) Florian Jetter
- Restore log message about received signals in CLI (:pr:`6618`) Florian Jetter
- Handle empty memoryviews of bytearrays when (de)serializing (:pr:`6576`) Benjamin Zaitlen
- Ensure steal requests from same-IP but distinct workers are rejected (:pr:`6585`) Florian Jetter
- Fix
tls_(min|max)_
version having no effect on openssl 1.1.0g or lower (:pr:`6562`) Thomas Grainger - Fix
idle_timeout
and unxfail test (:pr:`6563`) Matthew Rocklin - Fix crashing debug statement in
_purge_state
(:pr:`6589`) crusaderky - Abort connections on
CancelledError
(:pr:`6574`) Thomas Grainger - Fix Active Memory Manager ignores
nbytes
thresholds (:pr:`6583`) crusaderky
- Deprecate
WorkerState
accessors (:pr:`6579`) crusaderky
- Remove ipython hack (:pr:`6599`) crusaderky
- Mypy enforce
--no-implicit-optional
(:pr:`6606`) crusaderky - Update versioneer: change from using
SafeConfigParser
toConfigParser
(:pr:`6605`) Thomas A Caswell - Warn unreachable for scheduler.py (:pr:`6611`) Florian Jetter
- Refactor
wait_for_state()
(:pr:`6581`) crusaderky - Hardcode
wait_for_signals
signal list (:pr:`6619`) Thomas Grainger - Always pick an open port when running tests (:pr:`6591`) Florian Jetter
- Log popen stdout/err when subprocess times out (:pr:`6567`) Gabe Joseph
- Fix
test_error_during_startup
(:pr:`6608`) Florian Jetter - Make
test_idle_timeout_no_workers
more robust (:pr:`6602`) Florian Jetter - Mypy enforce
--disallow-incomplete-defs
(:pr:`6601`) crusaderky - Do not log during signal handler (:pr:`6590`) Florian Jetter
- Don't initialize
mp_context
on import (:pr:`6580`) Lawrence Mitchell - Test retire workers deadlock (:pr:`6240`) Gabe Joseph
- Rework some tests related to
gather_dep
(:pr:`6472`) crusaderky - Minor cosmetic review of
scheduler_story
andworker_story
(:pr:`6442`) crusaderky - Force
__future__.annotations
with isort (:pr:`6621`) Thomas Grainger
Released on June 10, 2022
- Make disk access in system monitor configurable (:pr:`6537`) Matthew Rocklin
- Log and except errors on preload start (:pr:`6553`) Matthew Rocklin
- Fix
Scheduler.restart
logic (:pr:`6504`) Gabe Joseph - Don't heartbeat while
Worker
is closing (:pr:`6543`) Gabe Joseph - No longer retry
LocalCluster
onerrno.EADDRINUSE
(:pr:`6369`) Thomas Grainger - Don't invoke
log_event
from state machine (:pr:`6512`) crusaderky - Add config option to disable profiling and disable it in many tests per default (:pr:`6490`) Hendrik Makait
- Encapsulate
Worker.batched_stream.send()
(:pr:`6475`) crusaderky
refresh-who-has
can break the worker state machine (:pr:`6529`) crusaderky- Restart worker if it's unrecognized by scheduler (:pr:`6505`) Gabe Joseph
- Fix import error when
distributed.rmm.pool-size
is set (:pr:`6482`) KoyamaSohei
- Restore signature compatibility for
dask-gateway
(:pr:`6561`) Tom Augspurger - Deprecate the
io_loop
andloop
kwarg toServer
,Worker
, andNanny
(:pr:`6473`) Thomas Grainger - Deprecate the
loop
kwarg toScheduler
(:pr:`6443`) Thomas Grainger
- Fix typo in
.nthreads()
docstring example (:pr:`6545`) Pavithra Eswaramoorthy - Update docs theme for rebranding (:pr:`6495`) Sarah Charlotte Johnson
- Refactor
gather_dep
(:pr:`6388`) crusaderky - Fix flaky
test_gather_dep_one_worker_always_busy
(:pr:`6554`) crusaderky - Remove
missing-data
message (:pr:`6546`) crusaderky - Port
test_local.LocalTest
to pytest tests to allow use ofloop
fixture (:pr:`6523`) Thomas Grainger - Fix
test_quiet_client_close
(:pr:`6541`) Gabe Joseph - Use
default_initializer
inWorkerProcess
(:pr:`6534`) jakirkham - Avoid deadlocks in tests that use
popen
(:pr:`6483`) Gabe Joseph - Revert "Fix CLI Scheduler Tests (:pr:`6502`)" (:pr:`6547`) Gabe Joseph
- Update test report URL in summary message (:pr:`6532`) Gabe Joseph
- Update test report url (:pr:`6531`) Ian Rose
- Assert
AsyncProcess.set_exit_callback
is not called with a coroutine function (:pr:`6526`) Thomas Grainger - Typing and docstring for
Worker.close
(:pr:`6518`) Hendrik Makait - Fix CLI Scheduler Tests (:pr:`6502`) Benjamin Zaitlen
- Collect assertions in
test_as_current_is_thread_local
(:pr:`6520`) Thomas Grainger - Link test report from test results comment (:pr:`6524`) Hendrik Makait
- Ignore the return value of
signal.signal
(:pr:`6519`) Thomas Grainger - Refactor all event handlers (:pr:`6410`) crusaderky
- Fix dashboard favicon background (:pr:`6514`) Jacob Tomlinson
- Update dashboard logo (:pr:`6513`) Jacob Tomlinson
- Fix
test_stress_scatter_death
(:pr:`6404`) Florian Jetter - Remove
CrossFilter
widget (:pr:`6484`) crusaderky data_needed
exclusively contains tasks in fetch state (:pr:`6481`) crusaderky- Assert possible previous states (:pr:`6488`) Florian Jetter
@fail_hard
can kill the whole test suite; hide errors (:pr:`6474`) crusaderky- Assert that a fetch->cancelled->resumed->fetch cycle is impossible (:pr:`6460`) crusaderky
- Refactor busy workers reinsertion (:pr:`6379`) crusaderky
- Refactor
find_missing
andrefresh_who_has
(:pr:`6348`) crusaderky - Rename
test_collections.py
totest_dask_collections.py
(:pr:`6486`) crusaderky update_who_has
can remove workers (:pr:`6342`) crusaderky- Restructure
test_watch_requires_lock_to_run
to avoid flakes (:pr:`6469`) Hendrik Makait - Fix intermittent
test_profile_plot
failure (:pr:`6456`) Matthew Rocklin - Use
asyncio.run
to rungen_cluster
,gen_test
andcluster
(:pr:`6231`) Thomas Grainger - Improve tests that watch for subprocess logs (:pr:`6461`) Gabe Joseph
Released on May 26, 2022
- Add a lock to
distributed.profile
for better concurrency control (:pr:`6421`) Hendrik Makait - Send
SIGKILL
afterSIGTERM
when passing 95% memory (:pr:`6419`) crusaderky
- Log rather than raise exceptions in
preload.teardown()
(:pr:`6458`) Matthew Rocklin - Handle failing
plugin.close()
calls during scheduler shutdown (:pr:`6450`) Matthew Rocklin - Fix slicing bug in
ensure_memoryview
(:pr:`6449`) jakirkham - Generalize UCX errors on
connect()
and correct pytest fixtures (:pr:`6434`) Peter Andreas Entschev - Run cluster widget periodic callbacks on the correct event loop (:pr:`6444`) Thomas Grainger
- Disable
pytest-asyncio
if installed (:pr:`6436`) Jacob Tomlinson - Close client in sync test_actor tests (:pr:`6459`) Thomas Grainger
- Ignore
ServerSession.with_document_locked unawaited
(:pr:`6447`) Thomas Grainger - Remove
coverage
pin from Python 3.10 environment (:pr:`6439`) Thomas Grainger - Annotate
remove_worker
(:pr:`6441`) crusaderky - Update gpuCI
RAPIDS_VER
to22.08
(:pr:`6428`)
Released on May 24, 2022
- Add HTTP API to scheduler (:pr:`6270`) Matthew Murray
- Shuffle Service with Scheduler Logic (:pr:`6007`) Matthew Rocklin
- Follow-up on removing
report
andsafe
fromWorker.close
(:pr:`6423`) Gabe Joseph - Server close faster (:pr:`6415`) Florian Jetter
- Disable HTTP API by default (:pr:`6420`) Jacob Tomlinson
- Remove
report
andsafe
fromWorker.close
(:pr:`6363`) Florian Jetter - Allow deserialized plugins in
register_scheduler_plugin
(:pr:`6401`) Matthew Rocklin WorkerState
are different for different addresses (:pr:`6398`) Florian Jetter- Do not filter tasks before gathering data (:pr:`6371`) crusaderky
- Remove worker reconnect (:pr:`6361`) Gabe Joseph
- Add
SchedulerPlugin.log_event handler
(:pr:`6381`) Matthew Rocklin - Ensure occupancy tracking works as expected for long running tasks (:pr:`6351`) Florian Jetter
stimulus_id
for allInstructions
(:pr:`6347`) crusaderky- Refactor missing-data command (:pr:`6332`) crusaderky
- Add
idempotent
toregister_scheduler_plugin
client (:pr:`6328`) Alex Ford - Add option to specify a scheduler address for workers to use (:pr:`5944`) Enric Tejedor
- Remove stray
breakpoint
(:pr:`6417`) Thomas Grainger - Fix API JSON MIME type (:pr:`6397`) Jacob Tomlinson
- Remove wrong
assert
in handle compute (:pr:`6370`) Florian Jetter - Ensure multiple clients can cancel their key without interference (:pr:`6016`) Florian Jetter
- Fix
Nanny
shutdown assertion (:pr:`6357`) Gabe Joseph - Fix
fail_hard
for sync functions (:pr:`6269`) Gabe Joseph - Prevent infinite transition loops; more aggressive
validate_state()
(:pr:`6318`) crusaderky - Ensure cleanup of many GBs of spilled data on terminate (:pr:`6280`) crusaderky
- Fix
WORKER_ANY_RUNNING
regression (:pr:`6297`) Florian Jetter - Race conditions from fetch to compute while AMM requests replica (:pr:`6248`) Florian Jetter
- Ensure resumed tasks are not accidentally forgotten (:pr:`6217`) Florian Jetter
- Do not allow closing workers to be awaited again (:pr:`5910`) Florian Jetter
- Move
wait_for_signals
to private module and deprecatedistributed.cli.utils
(:pr:`6367`) Hendrik Makait
- Fix typos and whitespace in
worker.py
(:pr:`6326`) Hendrik Makait - Fix link to memory trimming documentation (:pr:`6317`) Marco Wolsza
- Make
gen_test
show up in VSCode test discovery (:pr:`6424`) Gabe Joseph - WSMR /
deserialize_task
(:pr:`6411`) crusaderky - Restore signal handlers after wait for signals is done (:pr:`6400`) Thomas Grainger
fail_hard
should reraise (:pr:`6399`) crusaderky- Revisit tests mocking
gather_dep
(:pr:`6385`) crusaderky - Fix flaky
test_in_flight_lost_after_resumed
(:pr:`6372`) Florian Jetter - Restore install_signal_handlers due to downstream dependencies (:pr:`6366`) Hendrik Makait
- Improve
catch_unhandled_exceptions
(:pr:`6358`) Gabe Joseph - Remove all invocations of
IOLoop.run_sync
from CLI (:pr:`6205`) Hendrik Makait - Remove
transition-counter-max
from config (:pr:`6349`) crusaderky - Use
list
comprehension inpickle_loads
(:pr:`6343`) jakirkham - Improve
ensure_memoryview
test coverage & make minor fixes (:pr:`6333`) jakirkham - Remove leaking reference to
workers
fromgen_cluster
(:pr:`6337`) Hendrik Makait - Partial annotations for
stealing.py
(:pr:`6338`) crusaderky - Validate and debug state machine on
handle_compute_task
(:pr:`6327`) crusaderky - Bump pyupgrade and clean up
# type: ignore
(:pr:`6293`) crusaderky gen_cluster
to write to/tmp
(:pr:`6335`) crusaderky- Transition table as a
ClassVar
(:pr:`6331`) crusaderky - Simplify
ensure_memoryview
test witharray
(:pr:`6322`) jakirkham - Refactor
ensure_communicating
(:pr:`6165`) crusaderky - Review scheduler annotations, part 2 (:pr:`6253`) crusaderky
- Use
w
forwriteable
branch inpickle_loads
(:pr:`6314`) jakirkham - Simplify frame handling in
ws
(:pr:`6294`) jakirkham - Use
ensure_bytes
fromdask.utils
(:pr:`6295`) jakirkham - Use
ensure_memoryview
inarray
deserialization (:pr:`6300`) jakirkham - Escape < > when generating Junit report (:pr:`6306`) crusaderky
- Use
codecs.decode
to deserialize errors (:pr:`6274`) jakirkham - Minimize copying in
maybe_compress
&byte_sample
(:pr:`6273`) jakirkham - Skip
test_release_evloop_while_spilling
on OSX (:pr:`6291`) Florian Jetter - Simplify logic in
get_default_compression
(:pr:`6260`) jakirkham - Cleanup old compression workarounds (:pr:`6259`) jakirkham
- Re-enable NVML monitoring for WSL (:pr:`6119`) Charles Blackmon-Luca
Released on May 2, 2022
This is a bugfix release for :issue:`this issue<6255>`.
- Handle
writeable
inbuffer_callback
(:pr:`6238`) jakirkham - Use
.data
with NumPy array allocation (:pr:`6242`) jakirkham
- Close executor in event loop if interpreter is closing (:pr:`6256`) Matthew Rocklin
Released on April 29, 2022
- Unblock event loop while waiting for
ThreadpoolExecutor
to shut down (:pr:`6091`) Florian Jetter RetireWorker
policy is done if removed (:pr:`6234`) Gabe Joseph- Pause to disable dependency gathering (:pr:`6195`) crusaderky
- Add
EOFError
to nannymultiprocessing.queue
except list (:pr:`6213`) Matthew Rocklin - Re-interpret error in lost worker scenario (:pr:`6193`) Matthew Rocklin
- Add Stimulus IDs to Scheduler (:pr:`6161`) Florian Jetter
- Set a five minute TTL for Dask workers (:pr:`6200`) Matthew Rocklin
- Add
distributed.metrics.monotonic
(:pr:`6181`) crusaderky - Send worker validation errors to scheduler and err on test completion (:pr:`6192`) Matthew Rocklin
- Redesign worker exponential backoff on busy-gather (:pr:`6173`) crusaderky
- Log all invalid worker transitions to scheduler (:pr:`6134`) Matthew Rocklin
- Make Graph dashboard plot have invisible axes (:pr:`6149`) Matthew Rocklin
- Remove
Nanny
auto_restart
state (:pr:`6138`) Matthew Rocklin
- Ensure scheduler events do not hold on to
TaskState
objects (:pr:`6226`) Florian Jetter - Allow pausing and choke event loop while spilling (:pr:`6189`) crusaderky
- Do not use UUID in stealing (:pr:`6179`) Florian Jetter
- Handle int worker names in info page (:pr:`6158`) Brett Naul
- Fix
psutil
dio counters none (:pr:`6093`) ungarj - Join
Nanny
watch thread (:pr:`6146`) Matthew Rocklin - Improve logging when closing workers (:pr:`6129`) Matthew Rocklin
- Avoid stack overflow in profiling (:pr:`6141`) Matthew Rocklin
- Clean up
SSHCluster
if failure to start (:pr:`6130`) Matthew Rocklin
- Deprecate
rpc
synchronous context manager (:pr:`6171`) Thomas Grainger
- Update
actors.rst
(:pr:`6167`) Scott Sievert
- Add
fail_hard
decorator for worker methods (:pr:`6210`) Matthew Rocklin - Do not require
pytest_timeout
(:pr:`6224`) Florian Jetter - Remove remaining
run_sync
calls from tests (:pr:`6196`) Thomas Grainger - Increase test timeout if debugger is running (:pr:`6218`) Florian Jetter
- Do not list closes keyword in list of bullet points (:pr:`6219`) Florian Jetter
- Harmonize (:pr:`6161`) and (:pr:`6173`) (:pr:`6207`) crusaderky
- Xfail
test_worker_death_timeout
(:pr:`6186`) Matthew Rocklin - Use random port in
test_dask_spec.py::test_text
(:pr:`6187`) Matthew Rocklin - Mark all websocket tests as flaky (:pr:`6188`) Matthew Rocklin
- Fix flaky
test_dont_steal_long_running_tasks
(:pr:`6197`) crusaderky - Cleanup names in stealing (:pr:`6185`) Matthew Rocklin
log_errors
decorator (:pr:`6184`) crusaderky- Pass
mypy
validation on Windows (:pr:`6180`) crusaderky - Add
locket
as a dependency instead of vendoring (:pr:`6166`) Michael Adkins - Remove unittestmock for
gather_dep
andget_data_from_worker
(:pr:`6172`) Florian Jetter mypy
tweaks (:pr:`6175`) crusaderky- Avoid easy deprecated calls to
asyncio.get_event_loop()
(:pr:`6170`) Thomas Grainger - Fix flaky
test_cancel_fire_and_forget
(:pr:`6099`) crusaderky - Remove deprecated code (:pr:`6144`) Matthew Rocklin
- Update link of test badge (:pr:`6154`) Florian Jetter
- Remove legacy state mappings (:pr:`6145`) Matthew Rocklin
- Fix
test_worker_waits_for_scheduler
(:pr:`6155`) Matthew Rocklin - Disallow leaked threads on windows (:pr:`6152`) Thomas Grainger
- Review annotations and docstrings in
scheduler.py
, part 1 (:pr:`6132`) crusaderky - Relax
test_asyncprocess.py::test_simple
(:pr:`6150`) Matthew Rocklin - Drop
cast
ing which is effectively a no-op (:pr:`6101`) jakirkham - Mark tests that use a specific port as flaky (:pr:`6139`) Matthew Rocklin
- AMM Suggestion namedtuples (:pr:`6108`) crusaderky
Released on April 15, 2022
- Add
KillWorker
Plugin (:pr:`6126`) Matthew Rocklin
- Sort worker list in info pages (:pr:`6135`) Matthew Rocklin
- Add back
Worker.transition_fetch_missing
(:pr:`6112`) Matthew Rocklin - Log state machine events (:pr:`6092`) crusaderky
- Migrate
ensure_executing
transitions to newWorkerState
event mechanism - part 1 (:pr:`6003`) crusaderky - Migrate
ensure_executing
transitions to newWorkerState
event mechanism - part 2 (:pr:`6062`) crusaderky - Annotate worker transitions to error (:pr:`6012`) crusaderky
- Avoid transitioning from memory/released to missing in worker (:pr:`6123`) Matthew Rocklin
- Don't try to reconnect client on interpreter shutdown (:pr:`6120`) Matthew Rocklin
- Wrap UCX init warnings in importable functions (:pr:`6121`) Charles Blackmon-Luca
- Cancel asyncio tasks on worker close (:pr:`6098`) crusaderky
- Avoid port collisions when defining port ranges (:pr:`6054`) crusaderky
- Avoid intermittent failure in
test_cancel_fire_and_forget
(:pr:`6131`) Matthew Rocklin - Ignore
bokeh
warning in pytest (:pr:`6127`) Matthew Rocklin - Start uncythonization (:pr:`6104`) Martin Durant
- Avoid redundant cleanup fixture in
gen_test
tests (:pr:`6118`) Thomas Grainger - Move
comm.close
to finally intest_comms
(:pr:`6109`) Florian Jetter - Use
async
withServer
intest_core.py
(:pr:`6100`) Matthew Rocklin - Elevate warnings to errors in the test suite (:pr:`6094`) Thomas Grainger
- Add
urllib3
to nightly conda builds (:pr:`6102`) James Bourbeau - Drop Blosc (:pr:`6027`) Matthew Rocklin
- Robust
test_get_returns_early
(:pr:`6090`) Florian Jetter - Overhaul
test_priorities.py
(:pr:`6077`) crusaderky - Remove
pytest-asyncio
(:pr:`6063`) Thomas Grainger - Clean up usage around plain
rpc
(:pr:`6082`) Florian Jetter - Drop OSX builds for Python 3.9 (:pr:`6073`) Florian Jetter
- Bump periods in
utils_test.wait_for
(:pr:`6081`) Florian Jetter - Check for ucx-py nightlies when updating gpuCI (:pr:`6006`) Charles Blackmon-Luca
- Type annotations for
profile.py
(:pr:`6067`) crusaderky - Fix flaky
test_worker_time_to_live
(:pr:`6061`) crusaderky - Fix flaky
test_as_completed_async_for_cancel
(:pr:`6072`) crusaderky - Fix regression in
test_weakref_cache
(:pr:`6033`) crusaderky - Trivial fix to
test_nanny_worker_port_range
(:pr:`6070`) crusaderky - Drop deprecated
tornado.netutil.ExecutorResolver
(:pr:`6031`) Thomas Grainger - Delete
asyncio.py
(:pr:`6066`) Thomas Grainger - Tweak conda environment files (:pr:`6037`) crusaderky
- Harden
test_abort_execution_to_fetch
and more (:pr:`6026`) crusaderky - Fix
test_as_completed_with_results_no_raise
and namecomm
(:pr:`6042`) Matthew Rocklin - Use more robust limits in
test_worker_memory
(:pr:`6055`) Florian Jetter
Released on April 1, 2022
Note
This is the first release with support for Python 3.10
- Add Python 3.10 support (:pr:`5952`) Thomas Grainger
- New cluster dump utilities (:pr:`5920`) Simon Perkins
- New
ClusterDump
SchedulerPlugin
for dumping cluster state on close (:pr:`5983`) Simon Perkins - Track Event Loop intervals in dashboard plot (:pr:`5964`) Matthew Rocklin
ToPickle
-Unpickle
on the Scheduler (:pr:`5728`) Mads R. B. Kristensen
- Retry on transient error codes in
preload
(:pr:`5982`) Matthew Rocklin - More idiomatic
mypy
configuration (:pr:`6022`) crusaderky - Name extensions and enable extension heartbeats (:pr:`5957`) Matthew Rocklin
- Better error message on misspelled executor annotation (:pr:`6009`) crusaderky
- Clarify that SchedulerPlugin must be subclassed (:pr:`6008`) crusaderky
- Remove duplication from stealing (:pr:`5787`) Duncan McGregor
- Remove cache in
iscoroutinefunction
to avoid holding on to refs (:pr:`5985`) Florian Jetter - Add title to individual plots (:pr:`5967`) Matthew Rocklin
- Specify average in timeseries titles (:pr:`5974`) Matthew Rocklin
- Do not catch
CancelledError
inCommPool
(:pr:`6005`) Florian Jetter
- Remove
distributed._ipython_utils
and dependents (:pr:`6036`) Thomas Grainger - Remove support for PyPy (:pr:`6029`) James Bourbeau
- Drop runtime dependency to setuptools (:pr:`6017`) crusaderky
- Remove heartbeats from events (:pr:`5989`) Matthew Rocklin
- Mention default value of Client's
timeout
(:pr:`5933`) Eric Engestrom - Update celery and other outdated 3rd party URLs (:pr:`5988`) Thomas Grainger
- Improve
test_hardware
test (:pr:`6039`) Matthew Rocklin - Short variant of test_report.html (:pr:`6034`) crusaderky
- Make
test_reconnect
async (:pr:`6000`) Matthew Rocklin - Update gpuCI
RAPIDS_VER
to22.06
(:pr:`5962`) - Add tiny test for
ToPickle
(:pr:`6021`) Matthew Rocklin - Remove
check_python_3
(broken withclick>=8.1.0
) (:pr:`6018`) Thomas Grainger - Fix black in CI (:pr:`6019`) crusaderky
- Add a hardware benchmark to test memory, disk, and network bandwidths (:pr:`5966`) Matthew Rocklin
- Relax variable
test_race
(:pr:`5993`) Matthew Rocklin - Skip
dask-ssh
tests withoutparamiko
(:pr:`5907`) Elliott Sales de Andrade - Remove
test_restart_sync_no_center
(:pr:`5994`) Matthew Rocklin - Set lower tick frequency in tests (:pr:`5977`) Matthew Rocklin
- Catch
NotADirectoryError
inSafeTemporaryDirectory
(:pr:`5984`) Florian Jetter - Fix flaky
test_weakref_cache
(:pr:`5978`) crusaderky - Fixup
test_worker_doesnt_await_task_completion
(:pr:`5979`) Matthew Rocklin - Use broader range in
test_nanny_worker_port_range
(:pr:`5980`) Matthew Rocklin - Use
tempfile
directory in clusterfixture
(:pr:`5825`) Florian Jetter - Drop
setuptools
fromdistributed
recipe (:pr:`5963`) jakirkham
Released on March 18, 2022
- Support dumping cluster state to URL (:pr:`5863`) Gabe Joseph
- Prevent data duplication on unspill (:pr:`5936`) crusaderky
- Encapsulate spill buffer and memory_monitor (:pr:`5904`) crusaderky
- Drop
pkg_resources
in favour ofimportlib.metadata
(:pr:`5923`) Thomas Grainger - Worker State Machine refactor: redesign
TaskState
and scheduler messages (:pr:`5922`) crusaderky - Tidying of OpenSSL 1.0.2/Python 3.9 (and earlier) handling (:pr:`5854`) jakirkham
zict
type annotations (:pr:`5905`) crusaderky- Add key to compute failed message (:pr:`5928`) Florian Jetter
- Change default log format to include timestamp (:pr:`5897`) Florian Jetter
- Improve type annotations in worker.py (:pr:`5814`) crusaderky
- Fix
progress_stream
teardown (:pr:`5823`) Thomas Grainger - Handle concurrent or failing handshakes in
InProcListener
(:pr:`5903`) Thomas Grainger - Make
log_event
threadsafe (:pr:`5946`) Gabe Joseph
- Fixes to documentation regarding plugins (:pr:`5940`) crendoncoiled
- Some updates to scheduling policies docs (:pr:`5911`) Gabe Joseph
- Fix
test_nanny_worker_port_range
hangs on Windows (:pr:`5956`) crusaderky - (REVERTED) Unblock event loop while waiting for ThreadpoolExecutor to shut down (:pr:`5883`) Florian Jetter
- Revert :pr:`5883` (:pr:`5961`) crusaderky
- Invert
event_name
check intest-report
job (:pr:`5959`) jakirkham - Only run
gh-pages
workflow ondask/distributed
(:pr:`5942`) jakirkham absolufy-imports
- No relative imports - PEP8 (:pr:`5924`) Florian Jetter- Fix
track_features
for distributed pre-releases (:pr:`5927`) Charles Blackmon-Luca - Xfail
test_submit_different_names
(:pr:`5916`) Florian Jetter - Fix
distributed
pre-release'sdistributed-impl
constraint (:pr:`5867`) Charles Blackmon-Luca - Mock process memory readings in test_worker.py (v2) (:pr:`5878`) crusaderky
- Drop unused
_round_robin
global variable (:pr:`5881`) jakirkham - Add GitHub URL for PyPi (:pr:`5886`) Andrii Oriekhov
- Mark
xfail
COMPILED testsskipif
instead (:pr:`5884`) Florian Jetter
Released on February 25, 2022
- Add the ability for
Client
to runpreload
code (:pr:`5773`) Bryan W. Weber
- Optionally use NumPy to allocate buffers (:pr:`5750`) jakirkham
- Add git hash to
distributed-impl
version (:pr:`5865`) Charles Blackmon-Luca - Immediately raise exception when trying to connect to a closed cluster (:pr:`5855`) Florian Jetter
- Lazily get
dask
version information (:pr:`5822`) Thomas Grainger - Remove the requirements to add
comm
to every handler (:pr:`5820`) Florian Jetter - Raise on unclosed comms in
check_instances
(:pr:`5836`) Florian Jetter - Python 3.8 f-strings (:pr:`5828`) crusaderky
- Constrained spill (:pr:`5543`) Naty Clementi
- Measure actual spilled bytes, not output of
sizeof()
(:pr:`5805`) crusaderky - Remove redundant
str()
conversions (:pr:`5810`) crusaderky - Cluster dump now excludes
run_spec
by default (:pr:`5803`) Florian Jetter - Dump more objects with
dump_cluster_state
(:pr:`5806`) crusaderky - Do not connect to any sockets on import (:pr:`5808`) Florian Jetter
- Avoid deadlock when two tasks are concurrently waiting for an unresolved
ActorFuture
(:pr:`5709`) Thomas Grainger
- Drop Python 3.7 (:pr:`5683`) James Bourbeau
- Remove support for UCX < 1.11.1 (:pr:`5859`) Peter Andreas Entschev
- Fix typo in memory types documentation relative links (:pr:`5845`) James Bourbeau
- Document and test spill->target hysteresis cycle (:pr:`5813`) crusaderky
- Fix flaky
test_remove_replicas_while_computing
(:pr:`5860`) crusaderky - Fix time based
test_assert_worker_story_malformed_story
parameterize (:pr:`5856`) Thomas Grainger - Remove
xfail
fromtest_no_unnecessary_imports_on_worker
(:pr:`5862`) crusaderky - Start building pre-releases with cythonized scheduler (:pr:`5831`) Charles Blackmon-Luca
- Do not mark tests
xfail
if they don't come up in time (:pr:`5824`) Florian Jetter - Use
gen_cluster
where possible intest_dask_worker.py
(:pr:`5842`) Florian Jetter - Generate junit report when
pytest-timeout
killspytest
(:pr:`5832`) crusaderky - Decrease timeout-minutes for GHA jobs (:pr:`5837`) Florian Jetter
- Fix some timeouts (:pr:`5647`) Florian Jetter
- Bump pre-release version to be greater than stable releases (:pr:`5816`) Charles Blackmon-Luca
- Do not run schedule jobs on forks (:pr:`5821`) Florian Jetter
- Remove
pillow<9
pin in CI (:pr:`5775`) Thomas Grainger - Show scheduled test runs in report (:pr:`5812`) Ian Rose
- Add obvious exclusions with pragma statement (:pr:`5801`) Sarah Charlotte Johnson
- Add coverage exclusions for cli files (:pr:`5800`) Sarah Charlotte Johnson
- Add pragma statements (:pr:`5749`) Sarah Charlotte Johnson
- Remove pragma: no cover from
distributed.cli.dask_ssh
(:pr:`5809`) Thomas Grainger - Add pragma - worker.py, client.py, stealing.py (:pr:`5827`) Sarah Charlotte Johnson
- Relax
distributed
/dask-core
dependencies for pre-releases (:pr:`5802`) Charles Blackmon-Luca - Remove
test_ucx_config_w_env_var
flaky condition (:pr:`5765`) Peter Andreas Entschev
Released on February 11, 2022
Note
This is the last release with support for Python 3.7
- Update
client.scheduler_info
inwait_for_workers
(:pr:`5785`) Matthew Rocklin - Increase robustness to
TimeoutError
during connect (:pr:`5096`) Florian Jetter - Respect
KeyboardInterrupt
insync
(:pr:`5758`) Thomas Grainger - Add workflow / recipe to generate Dask/distributed pre-releases (:pr:`5636`) Charles Blackmon-Luca
- Review
Scheduler
/Worker
display repr (:pr:`5746`) crusaderky - AMM: Graceful Worker Retirement (:pr:`5381`) crusaderky
- AMM: tentatively stabilize flaky tests around worker pause (:pr:`5735`) crusaderky
- AMM: speed up and stabilize test_memory (:pr:`5737`) crusaderky
- Defer pandas import on worker in P2P shuffle (:pr:`5695`) Gabe Joseph
- Fix for
distributed.worker.memory.target=False
andspill=0.7
(:pr:`5788`) crusaderky - Transition
flight
tomissing
if nowho_has
(:pr:`5653`) Florian Jetter
- Remove deprecated
ncores
(:pr:`5780`) crusaderky - Deprecate registering plugins by class (:pr:`5699`) Thomas Grainger
- Deprecate
--nprocs
option fordask-worker
CLI (:pr:`5641`) Bryan W. Weber
- Fix imbalanced backticks (:pr:`5784`) Matthias Bussonnier
- xfail
test_worker_reconnects_mid_compute
(:pr:`5797`) crusaderky - Fix linting CI build (:pr:`5794`) James Bourbeau
- Update
pre-commit
versions (:pr:`5782`) James Bourbeau - Reactivate
pytest_resourceleaks
(:pr:`5771`) crusaderky - Set test assumption for
test_client_timeout
(:pr:`5790`) Florian Jetter - Remove client timeout from
test_ucx_config_w_env_var
(:pr:`5792`) Florian Jetter - Remove
test_failed_worker_without_warning
(:pr:`5789`) Florian Jetter - Fix longitudinal report (:pr:`5783`) Ian Rose
- Fix flaky
test_robust_to_bad_sizeof_estimates
(:pr:`5753`) crusaderky - Revert "Pin coverage to 6.2 (:pr:`5716`)" (:pr:`5770`) Thomas Grainger
- Trigger test runs periodically to increases failure statistics (:pr:`5769`) Florian Jetter
- More fault tolerant test report (:pr:`5732`) Ian Rose
- Pin
pillow<9
to work aroundtorch
incompatibility (:pr:`5755`) Thomas Grainger - Overhaul
check_process_leak
(:pr:`5739`) crusaderky - Fix flaky
test_exit_callback test
(:pr:`5713`) Jim Crist-Harif - Generate tests summary (:pr:`5710`) crusaderky
- Upload different architectured pre-releases separately (:pr:`5741`) Charles Blackmon-Luca
- Ignore non-test directories (:pr:`5720`) Gabe Joseph
- Bump gpuCI
PYTHON_VER
to 3.9 (:pr:`5738`) Charles Blackmon-Luca - Regression: threads noted down before they start (:pr:`5796`) crusaderky
Released on January 28, 2022
- P2P shuffle skeleton (:pr:`5520`) Gabe Joseph
- Fix
<Task pending name='...' coro=<Client._handle_report()>
(:pr:`5721`) Thomas Grainger - Add
distributed.client.security-loader
config (:pr:`5693`) Jim Crist-Harif - Avoid
Client._handle_report
cancelling itself onClient._close
(:pr:`5672`) Thomas Grainger - Paused workers shouldn't steal tasks (:pr:`5665`) crusaderky
- Add option for timestamps from output of
Node.get_logs
(:pr:`4932`) Charles Blackmon-Luca - Don't use
time.time()
orIOLoop.time()
(:pr:`5661`) crusaderky
- Raise plugin exceptions on
Worker.start()
(:pr:`4298`) Peter Andreas Entschev
- Fixing docstrings (:pr:`5696`) Julia Signell
- Fix typo in
Client.run
docstring (:pr:`5687`) Thomas Grainger - Update
client.py
docstrings (:pr:`5670`) Tim Harris
- Skip shuffle tests if
pandas
/dask.dataframe
not installed (:pr:`5730`) James Bourbeau - Improve test coverage (:pr:`5655`) Sarah Charlotte Johnson
- Test report improvements (:pr:`5714`) Ian Rose
- P2P shuffle: ignore row order in tests (:pr:`5706`) Gabe Joseph
- Fix flaky
test_no_reconnect[--no-nanny]
(:pr:`5686`) Thomas Grainger - Pin coverage to 6.2 (:pr:`5716`) Thomas Grainger
- Check for new name of timeouts artifact and be more fault tolerant (:pr:`5707`) Ian Rose
- Revisit rebalance unit tests (:pr:`5697`) crusaderky
- Update comment in
rearrange_by_column_p2p
(:pr:`5701`) James Bourbeau - Update gpuCI
RAPIDS_VER
to22.04
(:pr:`5676`) - Fix groupby test after meta requirements got stricter in Dask PR#8563 (:pr:`5694`) Julia Signell
- Fix flaky
test_close_gracefully
andtest_lifetime
(:pr:`5677`) crusaderky - Fix flaky
test_workspace_concurrency
(:pr:`5690`) crusaderky - Fix flaky
test_shuffle_extension.py::test_get_partition
(:pr:`5689`) Gabe Joseph - Fix flaky
test_dump_cluster_unresponsive_remote_worker
(:pr:`5679`) crusaderky - Dump cluster state on all test failures (:pr:`5674`) crusaderky
- Update license format (:pr:`5652`) James Bourbeau
- Fix flaky
test_drop_with_paused_workers_with_running_tasks_3_4
(:pr:`5673`) crusaderky - Do not raise an exception if the GitHub token cannot be found (:pr:`5668`) Florian Jetter
Released on January 14, 2022
- Task group stacked area chart (:pr:`5320`) Ian Rose
- Support configuring TLS min/max version (:pr:`5594`) Jim Crist-Harif
- Use asyncio for TCP/TLS comms (:pr:`5450`) Jim Crist-Harif
- Close comm on
CancelledError
(:pr:`5656`) crusaderky - Don't drop from the only running worker (:pr:`5626`) crusaderky
- Transfer priority (:pr:`5625`) crusaderky
- Add RPC call for getting task prefixes (:pr:`5617`) Benjamin Zaitlen
- Long running occupancy (:pr:`5395`) Florian Jetter
- Handle errors on individual workers in
run
/broadcast
(:pr:`5590`) crusaderky - Allow work stealing in case there are heterogeneous resources for thief and victim (:pr:`5573`) Florian Jetter
- Disable NVML monitoring on WSL (:pr:`5568`) Charles Blackmon-Luca
- Ensure uniqueness of steal stimulus ID (:pr:`5620`) Florian Jetter
- Fix
KeyError: 'startstops'
in performance report (:pr:`5608`) Gabe Joseph - Story timestamps can be slightly in the future (:pr:`5612`) crusaderky
- Prevent
RecursionError
inWorker._to_dict
(:pr:`5591`) crusaderky - Ensure distributed can be imported in thread (:pr:`5593`) Jim Crist-Harif
- Fix changelog section hyperlinks (:pr:`5638`) Aneesh Nema
- Fix typo in
unpublish_dataset
example invocation (:pr:`5615`) Deepyaman Datta - Fix typo in test report badge in
README
(:pr:`5586`) James Bourbeau
- Cosmetic changes to
distributed.comm
(:pr:`5657`) crusaderky - Consolidate broken comm testing utilities (:pr:`5654`) James Bourbeau
- Fix concurrency assumptions for
test_worker_reconnects_mid_compute
(:pr:`5623`) Florian Jetter - Handle Bokeh 3.0 CDSView change (:pr:`5643`) Bryan Van de Ven
- Use
packaging
rather thandistutils
to get version (:pr:`5624`) Julia Signell - XFAIL tls explicit comm close test on py3.7 (:pr:`5639`) Jim Crist-Harif
- Mark some additional ucx-py tests for GPU (:pr:`5603`) Charles Blackmon-Luca
- Rename
ensure_default_get
and add test (:pr:`5609`) Naty Clementi - Remove
render_mode
kwarg
fromboekh
LabelSets
(:pr:`5616`) Garry O'Donnell - Add lambda support to
assert_worker_story
(:pr:`5618`) crusaderky - Ignore file not found warning for timeout artifact (:pr:`5619`) Florian Jetter
- Improved cluster state dump in
@gen_cluster
(:pr:`5592`) crusaderky - Work around SSL failures on MacOS CI (:pr:`5606`) crusaderky
- Bump gpuCI
CUDA_VER
to 11.5 (:pr:`5604`) Charles Blackmon-Luca assert_worker_story
(:pr:`5598`) crusaderkydistributed.versions
code refresh (:pr:`5600`) crusaderky- Updates to gpuCI and
test_ucx_config_w_env_var
(:pr:`5595`) James Bourbeau - Replace blacklist/whitelist with blocklist/allowlist (:pr:`5589`) crusaderky
- Distributed test report (:pr:`5583`) Ian Rose
- AMM: cosmetic tweaks (:pr:`5584`) crusaderky
Released on December 10, 2021
- Support pytest fixures and parametrize with
gen_test
(:pr:`5532`) Fábio Rosado - Allow idempotent scheduler plugins to be registered via the RPC (:pr:`5545`) Jacob Tomlinson
- AMM logging (:pr:`5530`) crusaderky
- Raise error if
asyncssh
isn't installed when usingSSHCluster
(:pr:`5535`) Fábio Rosado - Allow
None
in UCX configuration schema (:pr:`5534`) Fábio Rosado - Add
distributed.comm.ucx.create-cuda-context
config (:pr:`5526`) Peter Andreas Entschev
- Allow unknown tasks to be stolen (:pr:`5572`) Florian Jetter
- Further
RecursionError
fixes inrecursive_to_repr
(:pr:`5579`) crusaderky - Revisit
recursive_to_dict
(:pr:`5557`) crusaderky - Handle
UCXUnreachable
exception (:pr:`5556`) Peter Andreas Entschev
- Separate
Coordination
section in API docs (:pr:`5412`) Gabe Joseph - Improved documentation for processing state and paused workers (:pr:`4985`) Maximilian Roos
- Fix typo in
TaskGroupGraph.update_layout
comment (:pr:`5536`) Hristo Georgiev - Update documentation for
register_worker_plugin
(:pr:`5533`) crusaderky
- Mark
test_gpu_monitoring_recent
as flaky (:pr:`5540`) Peter Andreas Entschev - Await worker arrival in SSH
test_nprocs
(:pr:`5575`) James Bourbeau - AMM: Test that acquire-replicas of a task already in flight is a no-op (:pr:`5566`) crusaderky
- Make sure artifacts are tagged with CI partition so they don't race and overwrite each other (:pr:`5571`) Ian Rose
- Minor refactoring and commentary in worker state machine (:pr:`5563`) Florian Jetter
- Fix
test_ucx_unreachable
on UCX < 1.12 (:pr:`5562`) Peter Andreas Entschev - Bump Bokeh min version to 2.1.1 (:pr:`5548`) Bryan Van de Ven
- Update
gen_test
tests to be more robust (:pr:`5551`) James Bourbeau - Skip
test_ucx_unreachable
ifUCXUnreachable
is unavailable (:pr:`5560`) Peter Andreas Entschev - Update gpuCI
RAPIDS_VER
to22.02
(:pr:`5544`) - Add workflow to automate gpuCI updates (:pr:`5541`) Charles Blackmon-Luca
- Actually support
uvloop
in distributed (:pr:`5531`) Jim Crist-Harif - Standardize UCX config separator to
-
(:pr:`5539`) Peter Andreas Entschev
Released on November 19, 2021
- Ensure cancelled error transition can properly release a key (:pr:`5528`) Florian Jetter
- Refactor release key (:pr:`5507`) Florian Jetter
- Fix deadlock caused by an erred task (executing->cancelled->error) (:pr:`5503`) Florian Jetter
- Resolve
KeyError
-related deadlock (:pr:`5525`) Florian Jetter - Remove extra quotation in worker failure docs (:pr:`5518`) James Bourbeau
- Ensure
safe_sizeof
warning is accurate (:pr:`5519`) James Bourbeau - Visualize cluster-wide memory usage over time (:pr:`5477`) crusaderky
- AMM: redesign start/stop methods (:pr:`5476`) crusaderky
- Preserve
contextvars
during comm offload (:pr:`5486`) Gabe Joseph - Deserialization: zero-copy merge subframes when possible (:pr:`5208`) Gabe Joseph
- Add support for multiple workers per SSH connection (:pr:`5506`) Jacob Tomlinson
- Client method to dump cluster state (:pr:`5470`) Florian Jetter
Released on November 8, 2021
- Revert "Avoid multiple blocking calls by gathering UCX frames" (:pr:`5505`) Peter Andreas Entschev
Released on November 5, 2021
- Fix
cluster_info
sync handling (:pr:`5488`) Jim Crist-Harif - Serialization family to preserve headers of the underlying dumps functions (:pr:`5380`) Mads R. B. Kristensen
- Point users to Discourse (:pr:`5489`) James Bourbeau
- Avoid multiple blocking calls by gathering UCX frames (:pr:`5487`) Peter Andreas Entschev
- Update all UCX tests to use
asyncio
marker (:pr:`5484`) Peter Andreas Entschev - Register UCX close callback (:pr:`5474`) Peter Andreas Entschev
- Use older version of
pynvml.nvmlDeviceGetComputeRunningProcesses
(:pr:`5469`) Jacob Tomlinson - Check for Futures from the wrong
Client
ingather
(:pr:`5468`) Gabe Joseph - Fix
performance_report
when used with%%time
or%%timeit
magic (:pr:`5463`) Erik Welch - Scatter and replicate to avoid paused workers (:pr:`5441`) crusaderky
- AMM to avoid paused workers (:pr:`5440`) crusaderky
- Update changelog with
LocalCluster
host security note (:pr:`5462`) Jim Crist-Harif
Released on October 22, 2021
Note
This release fixed a potential security vulnerability relating to
single-machine Dask clusters. Clusters started with
dask.distributed.LocalCluster
or dask.distributed.Client()
(which
defaults to using LocalCluster
) would mistakenly configure their
respective Dask workers to listen on external interfaces (typically with a
randomly selected high port) rather than only on localhost
. A Dask
cluster created using this method AND running on a machine that has these
ports exposed could be used by a sophisticated attacker to enable remote
code execution. Users running on machines with standard firewalls in place
should not be affected. This vulnerability is documented in CVE-2021-42343, and is fixed
in this release (:pr:`5427`). Thanks to Jean-Pierre van Riel for
discovering and reporting the issue.
- Ensure resumed flight tasks are still fetched (:pr:`5426`) Florian Jetter
- AMM high level documentation (:pr:`5456`) crusaderky
- Provide stack for suspended coro in test timeout (:pr:`5446`) Florian Jetter
- Handle
UCXNotConnected
error (:pr:`5449`) Peter Andreas Entschev - Don't schedule tasks to paused workers (:pr:`5431`) crusaderky
- Use
pip install .
instead of callingsetup.py
(:pr:`5442`) Matthias Bussonnier - Increase latency for stealing (:pr:`5390`) Florian Jetter
- Type annotations for
Worker
andgen_cluster
(:pr:`5438`) crusaderky - Ensure reconnecting workers do not loose required data (:pr:`5436`) Florian Jetter
- Mark
test_gather_dep*
asxfail
(:pr:`5432`) crusaderky - Remove
zict
-related skips (:pr:`5429`) James Bourbeau - Pass
host
throughLocalCluster
to workers (:pr:`5427`) Jim Crist-Harif - Fixes
async
warnings in UCX tests (:pr:`5396`) Peter Andreas Entschev - Resolve work stealing deadlock caused by race in
move_task_confirm
(:pr:`5379`) Florian Jetter - Add scroll to dashboard dropdown (:pr:`5418`) Jacob Tomlinson
- Fix regression where unknown tasks were allowed to be stolen (:pr:`5392`) Florian Jetter
- Enable
mypy
in CI 2/2 (:pr:`5348`) crusaderky - Rewrite
test_client_timeout
(:pr:`5397`) crusaderky - Simple
SSHCluster
example (:pr:`5349`) Ray Bell - Do not attempt to fetch keys which are no longer in flight (:pr:`5160`) Florian Jetter
- Revisit
Scheduler.add_plugin
/Scheduler.remove_plugin
(:pr:`5394`) crusaderky - Fix flaky
test_WorkerPlugin_overwrite
(:pr:`5398`) crusaderky - Active Memory Manager to use bulk comms (:pr:`5357`) crusaderky
- Add coverage badge to
README
(:pr:`5382`) James Bourbeau - Mark
test_stress_creation_and_deletion
asxfail
(:pr:`5393`) James Bourbeau - Mark
test_worker_reconnects_mid_compute*
tests as flaky (:pr:`5378`) James Bourbeau - Use new Dask docs theme (:pr:`5391`) Jacob Tomlinson
- Remove
pytest.mark.repeat
fromtest_prometheus_collect_task_states
(:pr:`5376`) James Bourbeau - Log original exception upon compute failure (:pr:`5387`) Florian Jetter
- Add code coverage (:pr:`4670`) James Bourbeau
- Fix zombie worker tasks after missing transition (:pr:`5316`) Florian Jetter
- Add support for partial functions to
iscoroutinefunction
util (:pr:`5344`) Michael Adkins - Mark
distributed/tests/test_client.py::test_profile_server
as flaky (:pr:`5375`) James Bourbeau - Enable
mypy
in CI 1/2 (:pr:`5328`) crusaderky - Ensure
dask-worker
anddask-scheduler
pick up preload configuration values (:pr:`5365`) James Bourbeau - Use
dask-spec
forSSHCluster
(:pr:`5191`) Charles Blackmon-Luca - Update
_cluster_info
dict in__init__
(:pr:`5305`) Jacob Tomlinson - Use Dask temporary file utility (:pr:`5361`) James Bourbeau
- Avoid deprecated random set sampling (:pr:`5360`) James Bourbeau
- Add check for unsupported NVML metrics (:pr:`5343`) Charles Blackmon-Luca
- Workers submit a reply to the scheduler if replica removal was rejected (:pr:`5356`) Florian Jetter
- Pickle exception and traceback immediately (:pr:`5338`) Mads R. B. Kristensen
- Reinstate: AMM
ReduceReplicas
to iterate only on replicated tasks (:pr:`5341`) crusaderky - Sync worker status to the scheduler; new 'paused' status (:pr:`5330`) crusaderky
- Add pre-commit to environments (:pr:`5362`) Ray Bell
- Worker State Machine Refactor: clean up dead handlers (:pr:`5359`) crusaderky
- Bump
RAPIDS_VER
for gpuCI (:pr:`5358`) Charles Blackmon-Luca - Generate Cython HTML annotations (:pr:`5321`) crusaderky
- Worker state machine refactor (:pr:`5046`) Florian Jetter
fsspec
ands3fs
git tips are incompatible (:pr:`5346`) crusaderky- Fix
test_many_Progress
and others (:pr:`5329`) crusaderky - Run multiple AMMs in parallel (:pr:`5339`) crusaderky
- Enhance AMM docstrings (:pr:`5340`) crusaderky
- Run
pyupgrade
in CI (:pr:`5327`) crusaderky - Fix typo in client side example
foundations.rst
(:pr:`5336`) Genevieve Buckley
Released on September 21, 2021
- Revert AMM
ReduceReplicas
and parallel AMMs updates (:pr:`5335`) James Bourbeau - Run multiple AMMs in parallel (:pr:`5315`) crusaderky
- AMM
ReduceReplicas
to iterate only on replicated tasks (:pr:`5297`) crusaderky - Add type annotations to various functions within
distributed.worker
(:pr:`5290`) Tom Forbes - Mark
test_ucx_config_w_env_var
flaky on UCX < 1.11 (:pr:`5262`) Peter Andreas Entschev - Warn if CUDA context is created on incorrect device in UCX (:pr:`5308`) Peter Andreas Entschev
- Remove redundant timeouts from
test_client
(:pr:`5314`) crusaderky - Allow
Client
to subscribe to events // Remote printing and warning (:pr:`5217`) Florian Jetter - Test pickle protocols 4 & 5 (:pr:`5313`) jakirkham
- Fix-up
test_pickle_empty
(:pr:`5303`) jakirkham - Increase timeout for
test_worker_reconnects_mid_compute_multiple_states_on_scheduler
(:pr:`5304`) Florian Jetter - Add synced dict between cluster and scheduler to store cluster info (:pr:`5033`) Jacob Tomlinson
- Update
test_sub_submit_priority
(:pr:`5301`) James Bourbeau - Revert "Add test setup fixture (:pr:`5242`)" (:pr:`5300`) James Bourbeau
- Fix flaky
test_worker_reconnects_mid_compute
(:pr:`5299`) Florian Jetter - Use
gen_test
intest_adaptive
(:pr:`5298`) crusaderky - Increase
worker.suspicious_counter
threshold (:pr:`5228`) Florian Jetter - Active Memory Manager framework + discard excess replicas (:pr:`5111`) crusaderky
- Add test setup fixture (:pr:`5242`) James Bourbeau
Released on September 3, 2021
- Fix
add_plugin
warnings (:pr:`5267`) Doug Davis - Add
list
around iterator inhandle_missing_dep
(:pr:`5285`) Matthew Rocklin - Jupyter-client 7 compatibility (:pr:`5286`) Min RK
- Replace
atop
withblockwise
(:pr:`5289`) James Bourbeau - Add pytest color to CI (:pr:`5276`) James Bourbeau
- Fix
test_map
and others (:pr:`5278`) crusaderky - Use
name
argument withScheduler.remove_plugin
calls (:pr:`5260`) Doug Davis - Downgrade to
jupyter_client
6 (:pr:`5273`) crusaderky - Migrate
Security
HTML repr to Jinja2 (:pr:`5264`) Jacob Tomlinson - Migrate
ProcessInterface
HTML repr to Jinja2 (:pr:`5263`) Jacob Tomlinson - Add support for diskless machines to system monitor (:pr:`5257`) James Bourbeau
- Avoid during-iteration scheduler plugin changes (:pr:`5259`) Doug Davis
- Remove
GroupProgress
scheduler plugin (:pr:`5256`) James Bourbeau - Properly check for ipv6 availability (:pr:`5255`) crusaderky
- Improved IPv6 dask-worker support (:pr:`5197`) Walt Woods
- Overwrite worker plugins (:pr:`5248`) Matthew Rocklin
- Refactor scheduler plugins; store in a dictionary (:pr:`5120`) Doug Davis
- Fix "then" -> "than" typo in docs (:pr:`5247`) David Chudzicki
- Fix typo (remove extra verb "creates") in docs (:pr:`5244`) David Chudzicki
- Fix "fractiom" -> "fraction" typo in docstring (:pr:`5245`) David Chudzicki
- Fix "schedulers" -> "scheduler" typo in docs (:pr:`5246`) David Chudzicki
- Use non-histogram plots up to 100 workers (:pr:`5249`) Matthew Rocklin
Released on August 20, 2021
- Rename plots to fit in the labextension (:pr:`5239`) Naty Clementi
- Log messages for
CommClosedError
now includes information about remote address (:pr:`5209`) Florian Jetter - Add
target='_blank'
for redirects of dashboard link (:pr:`5237`) Naty Clementi - Update computation code retrieval logic (:pr:`5236`) James Bourbeau
- Minor polish on cfexecutor (:pr:`5233`) crusaderky
- Use development version of
dask
in gpuCI build (:pr:`5232`) James Bourbeau - Use upstream
dask.widgets
(:pr:`5205`) Jacob Tomlinson - Fix flaky
test_worker_reconnects_mid_compute
(:pr:`5227`) Florian Jetter - Update
WorkerPlugin
docstring about usage ofTaskState
objects (:pr:`5226`) Florian Jetter - Worker Network Timeseries (:pr:`5129`) Naty Clementi
- Add HTML Repr for
ProcessInterface
class and all its subclasses (:pr:`5181`) Freyam Mehta - Fix an issue where a reconnecting worker could cause an invalid transition (:pr:`5210`) Florian Jetter
- Minor fixes for cfexecutor (:pr:`5177`) Florian Jetter
- Add HTML Repr for
Security
class (:pr:`5178`) Freyam Mehta - Fix performance report sizing issue (:pr:`5213`) Ian Rose
- Drop RMM compatibility code from RAPIDS < 0.11 (:pr:`5214`) Peter Andreas Entschev
Released on August 13, 2021
- Include addresses in closed comm repr (:pr:`5203`) James Bourbeau
- Test
nanny.environ
precedence (:pr:`5204`) Florian Jetter - Migrating HTML reprs to jinja2 (:pr:`5188`) Jacob Tomlinson
- Fix
test_process_executor_kills_process
flakyness (:pr:`5183`) crusaderky - Remove
urllib3
as a dependency downloading preloads (:pr:`5199`) Marcos Moyano - Download preload urls in the
Preload
constructor (:pr:`5194`) Marcos Moyano - Avoid recursion error in
profile.merge
(:pr:`5195`) Matthew Rocklin - Add text exceptions to the
Scheduler
(:pr:`5148`) Matthew Rocklin - Use
kwarg
forTheme
filename (:pr:`5190`) Bryan Van de Ven - Add a
.git-ignore-revs
file (:pr:`5187`) Florian Jetter - Replace
not not
withbool()
(:pr:`5182`) Jacob Tomlinson - Resolve deadlock cause by transition error after fetching dependency (:pr:`5157`) Florian Jetter
- Set z-index of data-table lower (:pr:`5175`) Julia Signell
- Add
no-worker
-memory
transition to scheduler (:pr:`5174`) Florian Jetter - Deprecate worker plugin overwrite policy (:pr:`5146`) James Bourbeau
- Fix flaky tests in CI (:pr:`5168`) crusaderky
- Instructions for jemalloc with brew on macOS (:pr:`4996`) Gabe Joseph
- Bump
RAPIDS_VER
to 21.10 (:pr:`5165`) Charles Blackmon-Luca - Tweak verbiage around
async
functions (:pr:`5166`) crusaderky - Use Python 3
super()
calls (:pr:`5167`) crusaderky - Support asynchronous tasks (:pr:`5151`) Matthew Rocklin
- Rename total comm bytes and provide doc string (:pr:`5155`) Florian Jetter
- Add GPU executor if GPU is present (:pr:`5123`) Matthew Rocklin
- Fix RMM and UCX tests (:pr:`5158`) Peter Andreas Entschev
- Remove excessive timeout of
test_steal_during_task_deserialization
(:pr:`5156`) Florian Jetter - Add gpuCI build script (:pr:`5147`) Charles Blackmon-Luca
- Demote
Worker.ensure_computing
to function (:pr:`5153`) Florian Jetter
Released on July 30, 2021
- Fix a deadlock connected to task stealing and task deserialization (:pr:`5128`) Florian Jetter
- Include maximum shard size in second
to_frames
method (:pr:`5145`) Matthew Rocklin - Minor dashboard style updates (:pr:`5143`) Bryan Van de Ven
- Cap maximum shard size at the size of an integer (:pr:`5141`) Matthew Rocklin
- Document automatic
MALLOC_TRIM_THRESHOLD_
environment variable (:pr:`5139`) James Bourbeau - Mark
ucx-py
tests for GPU (:pr:`5133`) Charles Blackmon-Luca - Update individual profile plot sizing (:pr:`5131`) James Bourbeau
- Handle
NVMLError_Unknown
in NVML diagnostics (:pr:`5121`) Peter Andreas Entschev - Unit tests to use a random port for the dashboard (:pr:`5060`) crusaderky
- Ensure worker reconnect registers existing tasks properly (:pr:`5103`) Florian Jetter
- Halve CI runtime! (:pr:`5074`) crusaderky
- Add
NannyPlugins
(:pr:`5118`) Matthew Rocklin - Add
WorkerNetworkBandwidth
chart to dashboard (:pr:`5104`) Naty Clementi - Set nanny environment variables in config (:pr:`5098`) Matthew Rocklin
- Read smaller frames to workaround OpenSSL bug (:pr:`5115`) jakirkham
- Move UCX/RMM config variables to Distributed namespace (:pr:`4916`) Charles Blackmon-Luca
- Allow ws(s) messages greater than 10Mb (:pr:`5110`) Marcos Moyano
- Short-circuit root-ish check for many deps (:pr:`5113`) Gabe Joseph
Released on July 23, 2021
- Remove experimental feature warning from actors docs (:pr:`5108`) James Bourbeau
- Keep dependents in worker dependency if TS is still known (:pr:`5034`) Florian Jetter
- Add
Scheduler.set_restrictions
(:pr:`5101`) Matthew Rocklin - Make
Actor
futures awaitable and work withas_completed
(:pr:`5092`) Martin Durant - Simplify
test_secede_balances
(:pr:`5071`) Florian Jetter Computation
class (:pr:`5001`) Florian Jetter- Some light dashboard cleanup (:pr:`5102`) Bryan Van de Ven
- Don't package tests (:pr:`5054`) James Bourbeau
- Add pytest marker for GPU tests (:pr:`5023`) Charles Blackmon-Luca
- Actor: don't hold key references on workers (:pr:`4937`) Gabe Joseph
- Collapse nav to hamburger sooner (:pr:`5094`) Julia Signell
- Verify that actors survive pickling (:pr:`5086`) Matthew Rocklin
- Reenable UCX-Py tests that used to segfault (:pr:`5076`) Peter Andreas Entschev
- Better support
ProcessPoolExecutors
(:pr:`5063`) Matthew Rocklin - Simplify
test_worker_heartbeat_after_cancel
(:pr:`5067`) Florian Jetter - Avoid property validation in Bokeh (:pr:`5065`) Matthew Rocklin
- Reduce default websocket frame size and make configurable (:pr:`5070`) Ian Rose
- Disable pytest-timeout
SIGALARM
on MacOS (:pr:`5057`) crusaderky rebalance()
resilience to computations (:pr:`4968`) crusaderky- Improve CI stability (:pr:`5022`) crusaderky
- Ensure heartbeats after cancelation do not raise
KeyError
s (:pr:`5053`) Florian Jetter - Add more useful exception message on TLS cert mismatch (:pr:`5040`) Jacob Tomlinson
- Add bokeh
mode
parameter to performance reports (:pr:`5025`) James Bourbeau
Released on July 9, 2021
- Fix Nbytes jitter - less expensive (:pr:`5043`) Naty Clementi
- Use native GH actions cancel feature (:pr:`5037`) Florian Jetter
- Don't require workers to report to scheduler if scheduler shuts down (:pr:`5032`) Florian Jetter
- Add pandas to the list of checked packages for
client.get_versions()
(:pr:`5029`) Ian Rose - Move worker preload before scheduler address is set (:pr:`5024`) Matthew Rocklin
- Fix flaky
test_oversubscribing_leases
(:pr:`5030`) Florian Jetter - Update scheduling policy docs for #4967 (:pr:`5018`) Gabe Joseph
- Add echo handler to
Server
class (:pr:`5020`) Matthew Rocklin - Also include pngs when bundling package (:pr:`5016`) Ian Rose
- Remove duplicated dashboard panes (:pr:`5017`) Ian Rose
- Fix worker memory dashboard flickering (:pr:`4997`) Naty Clementi
- Tabs on bottom left corner on dashboard (:pr:`5006`) Naty Clementi
- Rename nbytes widgets (:pr:`4878`) crusaderky
- Co-assign root-ish tasks (:pr:`4967`) Gabe Joseph
OSError
tweaks (:pr:`5003`) crusaderky- Update imports to
cudf.testing._utils
(:pr:`5005`) Peter Andreas Entschev - Ensure shuffle split default durations uses proper prefix (:pr:`4991`) Florian Jetter
- Follow up
pyupgrade
formatting (:pr:`4993`) Florian Jetter - Rename plot dropdown (:pr:`4992`) James Bourbeau
- Pyupgrade (:pr:`4741`) Florian Jetter
- Misc Sphinx tweaks (:pr:`4988`) crusaderky
- No longer hold dependencies of erred tasks in memory #4918 Florian Jetter
- Add maximum shard size to config (:pr:`4986`) Matthew Rocklin
- Ensure shuffle split operations are blacklisted from work stealing (:pr:`4964`) Florian Jetter
- Add dropdown menu to access individual plots (:pr:`4984`) Jacob Tomlinson
- Edited the path to
scheduler.py
(:pr:`4983`) Freyam Mehta - Task Group Graph Visualization (:pr:`4886`) Naty Clementi
- Remove more internal references to deprecated utilities (:pr:`4971`) James Bourbeau
- Restructure nbytes hover (:pr:`4952`) Naty Clementi
- Except more errors in
pynvml.nvmlInit()
(:pr:`4970`) gerrymanoim - Add occupancy as individual plot (:pr:`4963`) Naty Clementi
- Deprecate utilities which have moved to dask (:pr:`4966`) James Bourbeau
- Ensure connectionpool does not leave comms if closed mid connect (:pr:`4951`) Florian Jetter
- Add support for registering scheduler plugins from Client (:pr:`4808`) Doug Davis
- Stealing dashboard fixes (:pr:`4948`) Florian Jetter
- Allow requirements verification to be ignored when loading backends from entrypoints (:pr:`4961`) Florian Jetter
- Add
Log
andLogs
to API docs (:pr:`4946`) James Bourbeau - Support fixtures and
pytest.mark.parametrize
withgen_cluster
(:pr:`4958`) Gabe Joseph
Released on June 22, 2021
- Revert refactor to
utils.Log[s]
andCluster.get_logs
(:pr:`4941`) Charles Blackmon-Luca - Use deprecation utility from Dask (:pr:`4924`) James Bourbeau
- Add transition counter to
Scheduler
(:pr:`4934`) Matthew Rocklin - Remove
nbytes_in_memory
(:pr:`4930`) Matthew Rocklin
Released on June 18, 2021
- Fix deadlock in
handle_missing_dep
if additional replicas are available (:pr:`4929`) Florian Jetter - Add configuration to enable/disable NVML diagnostics (:pr:`4893`) Peter Andreas Entschev
- Add scheduler log tab to performance reports (:pr:`4909`) Charles Blackmon-Luca
- Add HTML repr to
scheduler_info
and incorporate into client and cluster reprs (:pr:`4857`) Jacob Tomlinson - Fix error state typo (:pr:`4898`) James Bourbeau
- Allow actor exceptions to propagate (:pr:`4232`) Martin Durant
- Remove importing
apply
fromdask.compatibility
(:pr:`4913`) Elliott Sales de Andrade - Use more informative default name for
WorkerPlugin
s (:pr:`4908`) James Bourbeau - Removed unused utility functions (:pr:`4911`) James Bourbeau
- Locally rerun successfully completed futures (:pr:`4813`) ArtinSarraf
- Forget erred tasks and fix deadlocks on worker (:pr:`4784`) Florian Jetter
- Handle
HTTPClientError
in websocket connector (:pr:`4900`) Marcos Moyano - Update
dask_cuda
usage inSSHCluster
docstring (:pr:`4894`) James Bourbeau - Remove tests for
process_time
andthread_time
(:pr:`4895`) James Bourbeau - Flake8 config cleanup (:pr:`4888`) Florian Jetter
- Don't strip scheduler protocol when determining host (:pr:`4883`) James Bourbeau
- Add more documentation on memory management (:pr:`4874`) crusaderky
- Add
range_query
tests to NVML test suite (:pr:`4879`) Charles Blackmon-Luca - No longer cancel result future in async process when using timeouts (:pr:`4882`) Florian Jetter
Released on June 4, 2021
- Multiple worker executors (:pr:`4869`) Mads R. B. Kristensen
- Ensure PyNVML works correctly when installed with no GPUs (:pr:`4873`) Peter Andreas Entschev
- Show more in test summary (:pr:`4875`) James Bourbeau
- Move
SystemMonitor
s GPU initialization back to constructor (:pr:`4866`) Peter Andreas Entschev - Mark
test_server_comms_mark_active_handlers
withpytest.mark.asyncio
(:pr:`4876`) James Bourbeau - Who has has what html reprs v2 (:pr:`4865`) Jacob Tomlinson
- O(1) rebalance (:pr:`4774`) crusaderky
- Ensure repr and eq for cluster always works (:pr:`4799`) Florian Jetter
Released on May 28, 2021
- Drop usage of
WhoHas
&WhatHas
fromClient
(:pr:`4863`) jakirkham - Ensure adaptive scaling is properly awaited and closed (:pr:`4720`) Florian Jetter
- Fix
WhoHas
/HasWhat
async
usage (:pr:`4860`) Benjamin Zaitlen - Add HTML reprs for
Client.who_has
andClient.has_what
(:pr:`4853`) Jacob Tomlinson - Prevent accidentally starting multiple
Worker
s in the same process (:pr:`4852`) crusaderky - Add system tab to performance reports (:pr:`4561`) Charles Blackmon-Luca
- Let servers close faster if there are no active handlers (:pr:`4805`) Florian Jetter
- Fix UCX scrub config logging (:pr:`4850`) Peter Andreas Entschev
- Ensure worker clients are closed (:pr:`3921`) Florian Jetter
- Fix warning for attribute error when deleting a client (:pr:`4807`) Florian Jetter
- Ensure exceptions are raised if workers are incorrectly started (:pr:`4733`) Florian Jetter
- Update handling of UCX exceptions on endpoint closing (:pr:`4836`) Peter Andreas Entschev
- Ensure busy workloads properly look up
who_has
(:pr:`4793`) Florian Jetter - Check
distributed.scheduler.pickle
inScheduler.run_function
(:pr:`4838`) James Bourbeau - Add performance_report to API docs (:pr:`4840`) James Bourbeau
- Use
dict
_workers_dv
in unordered use cases (:pr:`4826`) jakirkham - Bump
pre-commit
hook versions (:pr:`4835`) James Bourbeau - Do not mindlessly spawn workers when no memory limit is set (:pr:`4397`) Torsten Wörtwein
test_memory
to usegen_cluster
(:pr:`4811`) crusaderky- Increase timeout of
gen_test
to 30s (:pr:`4821`) Florian Jetter
Released on May 14, 2021
- Merge global annotations on the client (:pr:`4691`) Mads R. B. Kristensen
- Add support for
click
8 (:pr:`4810`) James Bourbeau - Add HTML reprs to some scheduler classes (:pr:`4795`) James Bourbeau
- Use JupyterLab theme variables (:pr:`4796`) Ian Rose
- Allow the dashboard to run on multiple ports (:pr:`4786`) Jacob Tomlinson
- Remove
release_dep
fromWorkerPlugin
API (:pr:`4791`) James Bourbeau - Support for UCX 1.10+ (:pr:`4787`) Peter Andreas Entschev
- Reduce complexity of
test_gather_allow_worker_reconnect
(:pr:`4739`) Florian Jetter - Fix doctests in
utils.py
(:pr:`4785`) Jacob Tomlinson - Ensure deps are actually logged in worker (:pr:`4753`) Florian Jetter
- Add
stacklevel
keyword intoperformance_report()
to allow for selecting calling code to be displayed (:pr:`4777`) Nathan Danielsen - Unregister worker plugin (:pr:`4748`) Naty Clementi
- Fixes some pickling issues in the Cythonized
Scheduler
(:pr:`4768`) jakirkham - Improve graceful shutdown if nanny is involved (:pr:`4725`) Florian Jetter
- Update cythonization in CI (:pr:`4764`) James Bourbeau
- Use
contextlib.nullcontext
(:pr:`4763`) James Bourbeau - Cython fixes for
MemoryState
(:pr:`4761`) jakirkham - Fix errors in
check_thread_leak
(:pr:`4747`) James Bourbeau - Handle missing
key
case inreport_on_key
(:pr:`4755`) jakirkham - Drop temporary
set
variabless
(:pr:`4758`) jakirkham
Released on April 23, 2021
- Avoid
active_threads
changing size during iteration (:pr:`4729`) James Bourbeau - Fix
UnboundLocalError
inAdaptiveCore.adapt()
(:pr:`4731`) Anderson Banihirwe - Minor formatting updates for HTTP endpoints doc (:pr:`4736`) James Bourbeau
- Unit test for
metrics["memory"]=None
(:pr:`4727`) crusaderky - Enable configuration of prometheus metrics namespace (:pr:`4722`) Jacob Tomlinson
- Reintroduce
weight
function (:pr:`4723`) James Bourbeau - Add
ready->memory
to transitions in worker (:pr:`4728`) Gil Forsyth - Fix regressions in :pr:`4651` (:pr:`4719`) crusaderky
- Add descriptions for UCX config options (:pr:`4683`) Charles Blackmon-Luca
- Split RAM measure into dask keys/other old/other new (:pr:`4651`) crusaderky
- Fix
DeprecationWarning
on Python 3.9 (:pr:`4717`) George Sakkis - ipython causes
test_profile_nested_sizeof
crash on windows (:pr:`4713`) crusaderky - Add
iterate_collection
argument toserialize
(:pr:`4641`) Richard J Zamora - When closing
Server
, close all listeners (:pr:`4704`) Florian Jetter - Fix timeout in
client.restart
(:pr:`4690`) Matteo De Wint - Avoid repeatedly using the same worker on first task with quiet cluster (:pr:`4638`) Doug Davis
- Grab
func
forfinish
case only if used (:pr:`4702`) jakirkham - Remove hostname check in
test_dashboard
(:pr:`4706`) James Bourbeau - Faster
tests_semaphore::test_worker_dies
(:pr:`4703`) Florian Jetter - Clean up
test_dashboard
(:pr:`4700`) crusaderky - Add timing information to
TaskGroup
(:pr:`4671`) Matthew Rocklin - Remove
WSSConnector
TLS presence check (:pr:`4695`) Marcos Moyano - Fix typo and remove unused
time.time
import (:pr:`4689`) Hristo Georgiev - Don't initialize CUDA context in monitor (:pr:`4688`) Charles Blackmon-Luca
- Add support for extra conn args for HTTP protocols (:pr:`4682`) Marcos Moyano
- Adjust timings in
test_threadpoolworkers
(:pr:`4681`) Florian Jetter - Add GPU metrics to
SystemMonitor
(:pr:`4661`) Charles Blackmon-Luca - Removing
dumps_msgpack()
andloads_msgpack()
(:pr:`4677`) Mads R. B. Kristensen - Expose worker
SystemMonitor
s to scheduler via RPC (:pr:`4657`) Charles Blackmon-Luca
Released on April 2, 2021
- Fix un-merged frames (:pr:`4666`) Matthew Rocklin
- Add informative error message to install uvloop (:pr:`4664`) Matthew Rocklin
- Remove incorrect comment regarding default
LocalCluster
creation (:pr:`4660`) cameron16 - Treat empty/missing
writeable
as a no-op (:pr:`4659`) jakirkham - Avoid list mutation in
pickle_loads
(:pr:`4653`) Matthew Rocklin - Ignore
OSError
exception when scaling down (:pr:`4633`) Gerald - Add
isort
to pre-commit hooks, package resorting (:pr:`4647`) Charles Blackmon-Luca - Use powers-of-two when displaying RAM (:pr:`4649`) crusaderky
- Support Websocket communication protocols (:pr:`4396`) Marcos Moyano
scheduler.py
/worker.py
code cleanup (:pr:`4626`) crusaderky- Update out-of-date references to
config.yaml
(:pr:`4643`) Hristo Georgiev - Suppress
OSError
onSpecCluster
shutdown (:pr:`4567`) Jacob Tomlinson - Replace conda with mamba (:pr:`4585`) crusaderky
- Expand documentation on pure functions (:pr:`4644`) James Lamb
Released on March 26, 2021
- Add standalone dashboard page for GPU usage (:pr:`4556`) Jacob Tomlinson
- Handle
stream is None
case in TCP comm finalizer (:pr:`4631`) James Bourbeau - Include
LIST_PICKLE
in NumPy array serialization (:pr:`4632`) James Bourbeau - Rename annotation plugin in
test_highlevelgraph.py
(:pr:`4618`) James Bourbeau - UCX use
nbytes
instead oflen
(:pr:`4621`) Mads R. B. Kristensen - Skip NumPy and pandas tests if not importable (:pr:`4563`) Ben Greiner
- Remove
utils.shutting_down
in favor ofsys.is_finalizing
(:pr:`4624`) James Bourbeau - Handle
async
clients when closing (:pr:`4623`) Matthew Rocklin - Drop
log
fromremove_key_from_stealable
(:pr:`4609`) jakirkham - Introduce events log length config option (:pr:`4615`) Fabian Gebhart
- Upstream config serialization and inheritance (:pr:`4372`) Jacob Tomlinson
- Add check to scheduler creation in
SpecCluster
(:pr:`4605`) Jacob Tomlinson - Make length of events
deque
configurable (:pr:`4604`) Fabian Gebhart - Add explicit
fetch
state to workerTaskState
(:pr:`4470`) Gil Forsyth - Update
develop.rst
(:pr:`4603`) Florian Jetter pickle_loads()
: Handle emptymemoryview
(:pr:`4595`) Mads R. B. Kristensen- Switch documentation builds for PRs to readthedocs (:pr:`4599`) James Bourbeau
- Track frame sizes along with frames (:pr:`4593`) jakirkham
- Add support for a list of keys when using
batch_size
inclient.map
(:pr:`4592`) Sultan Orazbayev - If
SpecCluster
fails to start attempt to gracefully close out again (:pr:`4590`) Jacob Tomlinson - Multi-lock extension (:pr:`4503`) Mads R. B. Kristensen
- Update
PipInstall
plugin command (:pr:`4584`) James Bourbeau - IPython magics: remove deprecated
ioloop
workarounds (:pr:`4530`) Min RK - Add GitHub actions workflow to cancel duplicate builds (:pr:`4581`) James Bourbeau
- Remove outdated macOS build badge from
README
(:pr:`4576`) James Bourbeau - Dask master -> main (:pr:`4569`) Julia Signell
- Drop support for Python 3.6 (:pr:`4390`) James Bourbeau
- Add docstring for
dashboard_link
property (:pr:`4572`) Doug Davis - Change default branch from master to main (:pr:`4495`) Julia Signell
- Msgpack handles extract serialize (:pr:`4531`) Mads R. B. Kristensen
Released on March 5, 2021
Note
This is the first release with support for Python 3.9 and the last release with support for Python 3.6
tcp.write()
: castmemoryview
to byte itemsize (:pr:`4555`) Mads R. B. Kristensen- Refcount the
thread_state.asynchronous
flag (:pr:`4557`) Mads R. B. Kristensen - Python 3.9 (:pr:`4460`) crusaderky
- Better bokeh defaults for dashboard (:pr:`4554`) Benjamin Zaitlen
- Expose system monitor dashboard as individual plot for lab extension (:pr:`4540`) Jacob Tomlinson
- Pass on original temp dir from nanny to worker (:pr:`4549`) Martin Durant
- Serialize and split (:pr:`4541`) Mads R. B. Kristensen
- Use the new HLG pack/unpack API in Dask (:pr:`4489`) Mads R. B. Kristensen
- Handle annotations for culled tasks (:pr:`4544`) Tom Augspurger
- Make sphinx autosummary and autoclass consistent (:pr:`4367`) Casey Clements
- Move
_transition*
toSchedulerState
(:pr:`4545`) jakirkham - Migrate from travis to GitHub actions (:pr:`4504`) crusaderky
- Move
new_task
toSchedulerState
(:pr:`4527`) jakirkham - Batch more Scheduler sends (:pr:`4526`) jakirkham
transition_memory_released
andget_nbytes()
optimizations (:pr:`4516`) jakirkham- Pin
black
pre-commit (:pr:`4533`) James Bourbeau - Read & write all frames in one pass (:pr:`4506`) jakirkham
- Skip
stream.write
call for empty frames (:pr:`4507`) jakirkham - Prepend frame metadata header (:pr:`4505`) jakirkham
transition_processing_memory
optimizations, etc. (:pr:`4487`) jakirkham- Attempt to get client from worker in
Queue
andVariable
(:pr:`4490`) James Bourbeau - Use
main
branch forzict
(:pr:`4499`) jakirkham - Use a callback to close TCP Comms, rather than check every time (:pr:`4453`) Matthew Rocklin
Released on February 5, 2021
- Bump minimum Dask to 2021.02.0 (:pr:`4486`) James Bourbeau
- Update
TaskState
documentation about dependents attribute (:pr:`4440`) Florian Jetter - DOC: Autoreformat all functions docstrings (:pr:`4475`) Matthias Bussonnier
- Use cached version of
is_coroutine_function
in stream handling to (:pr:`4481`) Ian Rose - Optimize
transitions
(:pr:`4451`) jakirkham - Create
PULL_REQUEST_TEMPLATE.md
(:pr:`4476`) Ray Bell - DOC: typo, directives ends with 2 colons
::
(:pr:`4472`) Matthias Bussonnier - DOC: Proper numpydoc syntax for
distributed/protocol/*.py
(:pr:`4473`) Matthias Bussonnier - Update
pytest.skip
usage intest_server_listen
(:pr:`4467`) James Bourbeau - Unify annotations (:pr:`4406`) Ian Rose
- Added worker resources from config (:pr:`4456`) Tom Augspurger
- Fix var name in worker validation func (:pr:`4457`) Gil Forsyth
- Refactor
task_groups
&task_prefixes
(:pr:`4452`) jakirkham - Use
parent._tasks
inheartbeat
(:pr:`4450`) jakirkham - Refactor
SchedulerState
fromScheduler
(:pr:`4365`) jakirkham
Released on January 22, 2021
- Make system monitor interval configurable (:pr:`4447`) Matthew Rocklin
- Add
uvloop
config value (:pr:`4448`) Matthew Rocklin - Additional optimizations to stealing (:pr:`4445`) jakirkham
- Give clusters names (:pr:`4426`) Jacob Tomlinson
- Use worker comm pool in
Semaphore
(:pr:`4195`) Florian Jetter - Set
runspec
on all new tasks to avoid deadlocks (:pr:`4432`) Florian Jetter - Support
TaskState
objects in story methods (:pr:`4434`) Matthew Rocklin - Support missing event loop in
Client.asynchronous
(:pr:`4436`) Matthew Rocklin - Don't require network to inspect tests (:pr:`4433`) Matthew Rocklin
Released on January 15, 2021
- Add time started to scheduler info (:pr:`4425`) Jacob Tomlinson
- Log adaptive error (:pr:`4422`) Jacob Tomlinson
- Xfail normalization tests (:pr:`4411`) Jacob Tomlinson
- Use
dumps_msgpack
andloads_msgpack
when packing high level graphs (:pr:`4409`) Mads R. B. Kristensen - Add
nprocs
auto option todask-worker
CLI (:pr:`4377`) Jacob Tomlinson - Type annotation of
_reevaluate_occupancy_worker
(:pr:`4398`) jakirkham - Type
TaskGroup
inactive_states
(:pr:`4408`) jakirkham - Fix
test_as_current_is_thread_local
(:pr:`4402`) jakirkham - Use
list
comprehensions to bindTaskGroup
type (:pr:`4401`) jakirkham - Make tests pass after 2028 (:pr:`4403`) Bernhard M. Wiedemann
- Fix compilation warnings,
decide_worker
now a C func, stealing improvements (:pr:`4375`) jakirkham - Drop custom
__eq__
fromStatus
(:pr:`4270`) jakirkham test_performance_report
: skip without bokeh (:pr:`4388`) Bruno PaganiNanny
now respects dask settings from ctx mgr (:pr:`4378`) Florian Jetter- Better task duration estimates for outliers (:pr:`4213`) selshowk
- Dask internal inherit config (:pr:`4364`) Jacob Tomlinson
- Provide
setup.py
option to profile Cython code (:pr:`4362`) jakirkham - Optimizations of
*State
andTask*
objects and stealing (:pr:`4358`) jakirkham - Cast
SortedDict
s todict
s in a few key places & other minor changes (:pr:`4355`) jakirkham - Use task annotation priorities for user-level priorities (:pr:`4354`) James Bourbeau
- Added docs to highlevelgraph pack/unpack (:pr:`4352`) Mads R. B. Kristensen
- Optimizations in notable functions used by transitions (:pr:`4351`) jakirkham
- Silence exception when releasing futures on process shutdown (:pr:`4309`) Benjamin Zaitlen
Released on December 10, 2020
- Switched to CalVer for versioning scheme.
- The scheduler can now receives Dask
HighLevelGraph
s instead of raw dictionary task graphs. This allows for a much more efficient communication of task graphs from the client to the scheduler. - Added support for using custom
Layer
-level annotations likepriority
,retries
, etc. with thedask.annotations
context manager. - Updated minimum supported version of Dask to 2020.12.0.
- Added many type annotations and updates to allow for gradually Cythonizing the scheduler.
- Some common optimizations across transitions (:pr:`4348`) jakirkham
- Drop serialize extension (:pr:`4344`) jakirkham
- Log duplicate workers in scheduler (:pr:`4338`) Matthew Rocklin
- Annotation of some comm related methods in the
Scheduler
(:pr:`4341`) jakirkham - Optimize
assert
invalidate_waiting
(:pr:`4342`) jakirkham - Optimize
decide_worker
(:pr:`4332`) jakirkham - Store occupancy in
_reevaluate_occupancy_worker
(:pr:`4337`) jakirkham - Handle
WorkerState
memory_limit
ofNone
(:pr:`4335`) jakirkham - Use
bint
to annotate boolean attributes (:pr:`4334`) jakirkham - Optionally use offload executor in worker (:pr:`4307`) Matthew Rocklin
- Optimize
send_task_to_worker
(:pr:`4331`) jakirkham - Optimize
valid_workers
(:pr:`4329`) jakirkham - Store occupancy in
transition_waiting_processing
(:pr:`4330`) jakirkham - Optimize
get_comm_cost
(:pr:`4328`) jakirkham - Use
.pop(...)
to removekey
(:pr:`4327`) jakirkham - Use
operator.attrgetter
onWorkerState.address
(:pr:`4324`) jakirkham - Annotate
Task*
objects for Cythonization (:pr:`4302`) jakirkham - Ensure
retire_workers
alwaysreturn
adict
(:pr:`4323`) jakirkham - Some Cython fixes for
WorkerState
(:pr:`4321`) jakirkham - Optimize
WorkerState.__eq__
(:pr:`4320`) jakirkham - Swap order of
TaskGroup
andTaskPrefix
(:pr:`4319`) jakirkham - Check traceback object can be unpickled (:pr:`4299`) jakirkham
- Move
TaskGroup
&TaskPrefix
before TaskState (:pr:`4318`) jakirkham - Remove empty
test_highgraph.py
file (:pr:`4313`) James Bourbeau - Ensure that
retire_workers
returns adict
(:pr:`4315`) Matthew Rocklin - Annotate
WorkerState
for Cythonization (:pr:`4294`) jakirkham - Close
comm
on low-level errors (:pr:`4239`) jochen-ott-by - Coerce new
TaskState.nbytes
value toint
(:pr:`4311`) jakirkham - Remove offload
try
/except
forthread_name_prefix
keyword (:pr:`4308`) James Bourbeau - Fix
pip
install issue on CI (:pr:`4310`) jakirkham - Transmit
Layer
annotations to scheduler (:pr:`4279`) Simon Perkins - Ignores any compiled files generated by Cython (:pr:`4301`) jakirkham
- Protect against missing key in
get_metrics
(:pr:`4300`) Matthew Rocklin - Provide option to build Distributed with Cython (:pr:`4292`) jakirkham
- Set
WorkerState.processing
w/dict
inclean
(:pr:`4295`) jakirkham - Annotate
ClientState
for Cythonization (:pr:`4290`) jakirkham - Annotate
check_idle_saturated
for Cythonization (:pr:`4289`) jakirkham - Avoid flicker in
TaskStream
with "Scheduler is empty" message (:pr:`4284`) Matthew Rocklin - Make
gather_dep
robust to missing tasks (:pr:`4285`) Matthew Rocklin - Annotate
extract_serialize
(for Cythonization) (:pr:`4283`) jakirkham - Move
nbytes
from Worker's state toTaskState
(:pr:`4274`) Gil Forsyth - Drop extra type check in
_extract_serialize
(:pr:`4281`) jakirkham - Move Status to top-level import (:pr:`4280`) Matthew Rocklin
- Add
__hash__
and__eq__
forTaskState
(:pr:`4278`) jakirkham - Add
__hash__
and__eq__
forClientState
(:pr:`4276`) jakirkham - Collect
report
'sclient_key``s in a ``list
(:pr:`4275`) jakirkham - Precompute
hash
forWorkerState
(:pr:`4271`) jakirkham - Use
Status
Enum
inremove_worker
(:pr:`4269`) jakirkham - Add aggregated topic logs and
log_event
method (:pr:`4230`) James Bourbeau - Find the set of workers instead of their frequency (:pr:`4267`) jakirkham
- Use
set.update
to include othercomms
(:pr:`4268`) jakirkham - Support string timeouts in
sync
(:pr:`4266`) James Bourbeau - Use
dask.utils.stringify()
instead ofdistributed.utils.tokey()
(:pr:`4255`) Mads R. B. Kristensen - Use
.items()
to walk through keys and values (:pr:`4261`) jakirkham - Simplify frame length packing in TCP write (:pr:`4257`) jakirkham
- Comm/tcp listener: do not pass comm with failed handshake to
comm_handler
(:pr:`4240`) jochen-ott-by - Fuse steps in
extract_serialize
(:pr:`4254`) jakirkham - Drop
test_sklearn
(:pr:`4253`) jakirkham - Document task priority tie breaking (:pr:`4252`) James Bourbeau
__dask_distributed_pack__()
: client argument (:pr:`4248`) Mads R. B. Kristensen- Configurable timeouts for
worker_client
andget_client
(:pr:`4146`) GeethanjaliEswaran - Add dask/distributed versions to
performance_report
(:pr:`4249`) Matthew Rocklin - Update miniconda GitHub action (:pr:`4250`) James Bourbeau
- UCX closing ignore error (:pr:`4236`) Mads R. B. Kristensen
- Redirect to
dask-worker
cli documentation (:pr:`4247`) Timost - Upload file worker plugin (:pr:`4238`) Ian Rose
- Create dependency
TaskState
as needed ingather_dep
(:pr:`4241`) Gil Forsyth - Instantiate plugin if needed in
register_worker_plugin
(:pr:`4198`) Julia Signell - Allow actors to call actors on the same worker (:pr:`4225`) Martin Durant
- Special case profile thread in leaked thread check (:pr:`4229`) James Bourbeau
- Use
intersection()
on a set instead ofdict_keys
inupdate_graph
(:pr:`4227`) Mads R. B. Kristensen - Communicate
HighLevelGraphs
directly to theScheduler
(:pr:`4140`) Mads R. B. Kristensen - Add
get_task_metadata
context manager (:pr:`4216`) James Bourbeau - Task state logs and data fix (:pr:`4206`) Gil Forsyth
- Send active task durations from worker to scheduler (:pr:`4192`) James Bourbeau
- Fix state check in
test_close_gracefully
(:pr:`4203`) Gil Forsyth - Avoid materializing layers in
Client.compute()
(:pr:`4196`) Mads R. B. Kristensen - Add
TaskState
metadata (:pr:`4191`) James Bourbeau - Fix regression in task stealing for already released keys (:pr:`4182`) Florian Jetter
- Fix
_graph_to_futures
bug for futures-based dependencies (:pr:`4178`) Richard J Zamora - High level graph
dumps
/loads
support (:pr:`4174`) Mads R. B. Kristensen - Implement pass HighLevelGraphs through
_graph_to_futures
(:pr:`4139`) Mads R. B. Kristensen - Support
async
preload click commands (:pr:`4170`) James Bourbeau dask-worker
cli memory limit option doc fix (:pr:`4172`) marwan116- Add
TaskState
toworker.py
(:pr:`4107`) Gil Forsyth - Increase robustness of
Semaphore.release
(:pr:`4151`) Lucas Rademaker - Skip batched comm test win / tornado5 (:pr:`4166`) Tom Augspurger
- Set Zict buffer target to maxsize when
memory_target_fraction
isFalse
(:pr:`4156`) Krishan Bhasin - Add
PipInstall
WorkerPlugin
(:pr:`3216`) Matthew Rocklin - Log
KilledWorker
events in the scheduler (:pr:`4157`) Matthew Rocklin - Fix
test_gpu_metrics
failure (:pr:`4154`) jakirkham
- Pin
pytest-asyncio
version (:pr:`4212`) James Bourbeau - Replace
AsyncProcess
exit handler byweakref.finalize
(:pr:`4184`) Peter Andreas Entschev - Remove hard coded connect handshake timeouts (:pr:`4176`) Florian Jetter
- Support
SubgraphCallable
instr_graph()
(:pr:`4148`) Mads R. B. Kristensen - Handle exceptions in
BatchedSend
(:pr:`4135`) Tom Augspurger - Fix for missing
:
in autosummary docs (:pr:`4143`) Gil Forsyth - Limit GPU metrics to visible devices only (:pr:`3810`) Jacob Tomlinson
- Use
pandas.testing
(:pr:`4138`) jakirkham - Fix a few typos (:pr:`4131`) Pav A
- Return right away in
Cluster.close
if cluster is already closed (:pr:`4116`) Tom Rochette - Update async doc with example on
.compute()
vsclient.compute()
(:pr:`4137`) Benjamin Zaitlen - Correctly tear down
LoopRunner
inClient
(:pr:`4112`) Sergey Kozlov - Simplify
Client._graph_to_futures()
(:pr:`4127`) Mads R. B. Kristensen - Cleanup new exception traceback (:pr:`4125`) Krishan Bhasin
- Stop writing config files by default (:pr:`4123`) Matthew Rocklin
- Fix SSL
connection_args
forprogressbar
connect (:pr:`4122`) jennalc
- Fix registering a worker plugin with
name
arg (:pr:`4105`) Nick Evans - Support different
remote_python
paths on cluster nodes (:pr:`4085`) Abdulelah Bin Mahfoodh - Allow
RuntimeError
s when closing global clients (:pr:`4115`) Matthew Rocklin - Match
pre-commit
in dask (:pr:`4049`) Julia Signell - Update
super
usage (:pr:`4110`) Poruri Sai Rahul
- Add logging for adaptive start and stop (:pr:`4101`) Matthew Rocklin
- Don't close a nannied worker if it hasn't yet started (:pr:`4093`) Matthew Rocklin
- Respect timeouts when closing clients synchronously (:pr:`4096`) Matthew Rocklin
- Log when downloading a preload script (:pr:`4094`) Matthew Rocklin
dask-worker --nprocs
accepts negative values (:pr:`4089`) Dror Speiser- Support zero-worker clients (:pr:`4090`) Matthew Rocklin
- Exclude
fire-and-forget
client from metrics (:pr:`4078`) Tom Augspurger - Drop
Serialized.deserialize()
method (:pr:`4073`) jakirkham - Add
timeout=
keyword toClient.wait_for_workers
method (:pr:`4087`) Matthew Rocklin
- Update for black (:pr:`4081`) Tom Augspurger
- Provide informative error when connecting an older version of Dask (:pr:`4076`) Matthew Rocklin
- Simplify
pack_frames
(:pr:`4068`) jakirkham - Simplify
frame_split_size
(:pr:`4067`) jakirkham - Use
list.insert
to add prelude up front (:pr:`4066`) jakirkham - Graph helper text (:pr:`4064`) Julia Signell
- Graph dashboard: Reset container data if task number is too large (:pr:`4056`) Florian Jetter
- Ensure semaphore picks correct
IOLoop
for threadpool workers (:pr:`4060`) Florian Jetter - Add cluster log method (:pr:`4051`) Jacob Tomlinson
- Cleanup more exception tracebacks (:pr:`4054`) Krishan Bhasin
- Improve documentation of
scheduler.locks
options (:pr:`4062`) Florian Jetter
- Move toolbar to above and fix y axis (#4043) Julia Signell
- Make behavior clearer for how to get worker dashboard (#4047) Julia Signell
- Worker dashboard clean up (#4046) Julia Signell
- Add a default argument to the datasets and a possibility to override datasets (#4052) Nils Braun
- Discover HTTP endpoints (#3744) Martin Durant
- Tidy up exception traceback in TCP Comms (:pr:`4042`) Krishan Bhasin
- Angle on the x-axis labels (:pr:`4030`) Mathieu Dugré
- Always set RMM's strides in the
header
(:pr:`4039`) jakirkham - Fix documentation
upload_file
(:pr:`4038`) Roberto Panai - Update UCX tests for new handshake step (:pr:`4036`) jakirkham
- Add test for informative errors in serialization cases (:pr:`4029`) Matthew Rocklin
- Add compression, pickle protocol to comm contexts (:pr:`4019`) Matthew Rocklin
- Make GPU plots robust to not having GPUs (:pr:`4008`) Matthew Rocklin
- Update
PendingDeprecationWarning
with correct version number (:pr:`4025`) Matthias Bussonnier - Install PyTorch on CI (:pr:`4017`) jakirkham
- Try getting cluster
dashboard_link
before asking scheduler (:pr:`4018`) Matthew Rocklin - Ignore writeable frames with builtin
array
(:pr:`4016`) jakirkham - Just extend
frames2
byframes
(:pr:`4015`) jakirkham - Serialize builtin array (:pr:`4013`) jakirkham
- Use cuDF's
assert_eq
(:pr:`4014`) jakirkham - Clear function cache whenever we upload a new file (:pr:`3993`) Jack Xiaosong Xu
- Emit warning when assign/comparing string with
Status
Enum
(:pr:`3875`) Matthias Bussonnier - Track mutable frames (:pr:`4004`) jakirkham
- Improve
bytes
andbytearray
serialization (:pr:`4009`) jakirkham - Fix memory histogram values in dashboard (:pr:`4006`) Willi Rath
- Only call
frame_split_size
when there are frames (:pr:`3996`) jakirkham - Fix failing
test_bandwidth
(:pr:`3999`) jakirkham - Handle sum of memory percentage when
memory_limit
is 0 (:pr:`3984`) Julia Signell - Drop msgpack pre-0.5.2 compat code (:pr:`3977`) jakirkham
- Revert to localhost for local IP if no network available (:pr:`3991`) Matthew Rocklin
- Add missing backtick in inline directive. (:pr:`3988`) Matthias Bussonnier
- Warn when
threads_per_worker
is set to zero (:pr:`3986`) Julia Signell - Use
memoryview
inunpack_frames
(:pr:`3980`) jakirkham - Iterate over list of comms (:pr:`3959`) Matthew Rocklin
- Streamline
pack_frames
/unpack_frames
frames (:pr:`3973`) jakirkham - Always attempt to create
dask-worker-space
folder and continue if it exists (:pr:`3972`) Jendrik Jördening - Use
merge_frames
with host memory only (:pr:`3971`) jakirkham - Simplify
pack_frames_prelude
(:pr:`3961`) jakirkham - Use continuation prompt for proper example parsing (:pr:`3966`) Matthias Bussonnier
- Ensure writable frames (:pr:`3967`) jakirkham
- Fix data replication error (:pr:`3963`) Andrew Fulton
- Treat falsey local directory as
None
(:pr:`3964`) Tom Augspurger - Unpin
numpydoc
now that 1.1 is released (:pr:`3957`) Gil Forsyth - Error hard when Dask has mismatched versions or lz4 installed (:pr:`3936`) Matthew Rocklin
- Skip coercing to
bytes
inmerge_frames
(:pr:`3960`) jakirkham - UCX: reuse endpoints in order to fix NVLINK issue (:pr:`3953`) Mads R. B. Kristensen
- Optionally use
pickle5
(:pr:`3849`) jakirkham - Update time per task chart with filtering and pie (:pr:`3933`) Benjamin Zaitlen
- UCX: explicit shutdown message (:pr:`3950`) Mads R. B. Kristensen
- Avoid too aggressive retry of connections (:pr:`3944`) Matthias Bussonnier
- Parse timeouts in
Client.sync
(:pr:`3952`) Matthew Rocklin - Synchronize on non-trivial CUDA frame transmission (:pr:`3949`) jakirkham
- Serialize
memoryview
withshape
andformat
(:pr:`3947`) jakirkham - Move
scheduler_comm
intoCluster.__init__
(:pr:`3945`) Matthew Rocklin
- Link issue on using
async
withexecutor_submit
(:pr:`3939`) jakirkham - Make dashboard server listens on all IPs by default even when interface is set explicitly (:pr:`3941`) Loïc Estève
- Update logic for worker removal in check ttl (:pr:`3927`) Benjamin Zaitlen
- Close a created cluster quietly (:pr:`3935`) Matthew Rocklin
- Ensure
Worker.run*
handleskwargs
correctly (:pr:`3937`) jakirkham - Restore
Scheduler.time_started
for Dask Gateway (:pr:`3934`) Tom Augspurger - Fix exception handling in
_wait_until_connected
(:pr:`3912`) Alexander Clausen - Make local directory if it does not exist (:pr:`3928`) Matthew Rocklin
- Install vanilla status route if bokeh dependency is not satisfied (:pr:`3844`) joshreback
- Make
Worker.delete_data
sync (:pr:`3922`) Peter Andreas Entschev - Fix
ensure_bytes
import location (:pr:`3919`) jakirkham - Fix race condition in repeated calls to
cluster.adapt()
(:pr:`3915`) Jacob Tomlinson
- Notify worker plugins when a task is released (:pr:`3817`) Nick Evans
- Update heartbeat checks in scheduler (:pr:`3896`) Benjamin Zaitlen
- Make encryption default if
Security
is given arguments (:pr:`3887`) Matthew Rocklin - Show
cpu_fraction
on hover for dashboard workers circle plot. (:pr:`3906`) Loïc Estève - Prune virtual client on variable deletion (:pr:`3910`) Marco Neumann
- Fix total aggregated metrics in dashboard (:pr:`3897`) Loïc Estève
- Support Bokeh 2.1 (:pr:`3904`) Matthew Rocklin
- Update
related-work.rst
(:pr:`3889`) DomHudson - Skip
test_pid_file
in older versions of Python (:pr:`3888`) Matthew Rocklin - Replace
stream=
withcomm=
in handlers (:pr:`3860`) Julien Jerphanion - Check hosts for
None
value in SSH cluster. (:pr:`3883`) Matthias Bussonnier - Allow dictionaries in
security=
keywords (:pr:`3874`) Matthew Rocklin - Use pickle protocol 5 with NumPy object arrays (:pr:`3871`) jakirkham
- Cast any
frame
touint8
(same type asbytes
) (:pr:`3870`) jakirkham - Use
Enum
for worker, scheduler and nanny status. (:pr:`3853`) Matthias Bussonnier - Drop legacy
buffer_interface
assignment (:pr:`3869`) jakirkham - Drop old frame splitting in NumPy serialization (:pr:`3868`) jakirkham
- Drop no longer needed local
import pickle
(:pr:`3865`) jakirkham - Fix typo in
feed
's log message (:pr:`3867`) jakirkham - Tidy pickle (:pr:`3866`) jakirkham
- Handle empty times in task stream (:pr:`3862`) Benjamin Zaitlen
- Change
asyncssh
objects to sphinx references (:pr:`3861`) Jacob Tomlinson - Improve
SSHCluster
docstring forconnect_options
(:pr:`3859`) Jacob Tomlinson - Validate address parameter in client constructor (:pr:`3842`) joshreback
- Use
SpecCluster
name in worker names (:pr:`3855`) Loïc Estève - Allow async
add_worker
andremove_worker
plugin methods (:pr:`3847`) James Bourbeau
- Merge frames in
deserialize_bytes
(:pr:`3639`) John Kirkham - Allow
SSHCluster
to take a list ofconnect_options
(:pr:`3854`) Jacob Tomlinson - Add favicon to performance report (:pr:`3852`) Jacob Tomlinson
- Add dashboard plots for the amount of time spent per key and for transfer/serialization (:pr:`3792`) Benjamin Zaitlen
- Fix variable name in journey of a task documentation (:pr:`3840`) Matthias Bussonnier
- Fix typo in journey of a task doc (:pr:`3838`) James Bourbeau
- Register
dask_cudf
serializers (:pr:`3832`) John Kirkham - Fix key check in
rebalance
missing keys (:pr:`3834`) Jacob Tomlinson - Allow collection of partial profile information in case of exceptions (:pr:`3773`) Florian Jetter
- Record the time since the last run task on the scheduler (:pr:`3830`) Matthew Rocklin
- Set colour of
nbytes
pane based on thresholds (:pr:`3805`) Krishan Bhasin - Include total number of tasks in the performance report (:pr:`3822`) Abdulelah Bin Mahfoodh
- Allow to pass in task key strings in the worker restrictions (:pr:`3826`) Nils Braun
- Control de/ser offload (:pr:`3793`) Martin Durant
- Parse timeout parameters in
Variable
/Event
/Lock
to support text timeouts (:pr:`3825`) Nils Braun - Don't send empty dependencies (:pr:`3423`) Jakub Beránek
- Add distributed Dask
Event
that mimicsthreading.Event
(:pr:`3821`) Nils Braun - Enhance
VersionMismatchWarning
messages (:pr:`3786`) Abdulelah Bin Mahfoodh - Support Pickle's protocol 5 (:pr:`3784`) jakirkham
- Replace
utils.ignoring
withcontextlib.suppress
(:pr:`3819`) Nils Braun - Make re-creating conda environments from the CI output easier (:pr:`3816`) Lucas Rademaker
- Add prometheus metrics for semaphore (:pr:`3757`) Lucas Rademaker
- Fix worker plugin called with superseded transition (:pr:`3812`) Nick Evans
- Add retries to server listen (:pr:`3801`) Jacob Tomlinson
- Remove commented out lines from
scheduler.py
(:pr:`3803`) James Bourbeau - Fix
RuntimeWarning
for never awaited coroutine when usingdistributed.Semaphore
(:pr:`3713`) Florian Jetter - Fix profile thread leakage during test teardown on some platforms (:pr:`3795`) Florian Jetter
- Await self before handling comms (:pr:`3788`) Matthew Rocklin
- Fix typo in
Cluster
docstring (:pr:`3787`) Scott Sanderson
Client.get_dataset
to always createFutures
attached to itself (:pr:`3729`) crusaderky- Remove dev-requirements since it is unused (:pr:`3782`) Julia Signell
- Use bokeh column for
/system
instead of custom css (:pr:`3781`) Julia Signell - Attempt to fix
test_preload_remote_module
on windows (:pr:`3775`) James Bourbeau - Fix broadcast for TLS comms (:pr:`3766`) Florian Jetter
- Don't validate http preloads locally (:pr:`3768`) Rami Chowdhury
- Allow range of ports to be specified for
Workers
(:pr:`3704`) James Bourbeau - Add UCX support for RDMACM (:pr:`3759`) Peter Andreas Entschev
- Support web addresses in preload (:pr:`3755`) Matthew Rocklin
- Connect to dashboard when address provided (:pr:`3758`) Tom Augspurger
- Move
test_gpu_metrics test
(:pr:`3721`) Tom Augspurger - Nanny closing worker on
KeyboardInterrupt
(:pr:`3747`) Mads R. B. Kristensen - Replace
OrderedDict
withdict
in scheduler (:pr:`3740`) Matthew Rocklin - Fix exception handling typo (:pr:`3751`) Jonas Haag
- Ensure
BokehTornado
uses prefix (:pr:`3746`) James Bourbeau - Warn if cluster closes before starting (:pr:`3735`) Matthew Rocklin
- Memoryview serialisation (:pr:`3743`) Martin Durant
- Allows logging config under distributed key (:pr:`2952`) Dillon Niederhut
- Reinstate support for legacy
@gen_cluster
functions (:pr:`3738`) crusaderky - Relax NumPy requirement in UCX (:pr:`3731`) jakirkham
- Add Configuration Schema (:pr:`3696`) Matthew Rocklin
- Reuse CI scripts for local installation process (:pr:`3698`) crusaderky
- Use
PeriodicCallback
class from tornado (:pr:`3725`) James Bourbeau - Add
remote_python
option in ssh cmd (:pr:`3709`) Abdulelah Bin Mahfoodh - Configurable polling interval for cluster widget (:pr:`3723`) Julia Signell
- Fix copy-paste in docs (:pr:`3728`) Julia Signell
- Replace
gen.coroutine
with async-await in tests (:pr:`3706`) crusaderky - Fix flaky
test_oversubscribing_leases
(:pr:`3726`) Florian Jetter - Add
batch_size
toClient.map
(:pr:`3650`) Tom Augspurger - Adjust semaphore test timeouts (:pr:`3720`) Florian Jetter
- Dask-serialize dicts longer than five elements (:pr:`3689`) Richard J Zamora
- Force
threads_per_worker
(:pr:`3715`) crusaderky - Idempotent semaphore acquire with retries (:pr:`3690`) Florian Jetter
- Always use
readinto
in TCP (:pr:`3711`) jakirkham - Avoid
DeprecationWarning
from pandas (:pr:`3712`) Tom Augspurger - Allow modification of
distributed.comm.retry
at runtime (:pr:`3705`) Florian Jetter - Do not log an error on unset variable delete (:pr:`3652`) Jonathan J. Helmus
- Add
remote_python
keyword to the newSSHCluster
(:pr:`3701`) Abdulelah Bin Mahfoodh - Replace Example with Examples in docstrings (:pr:`3697`) Matthew Rocklin
- Add
Cluster
__enter__
and__exit__
methods (:pr:`3699`) Matthew Rocklin - Fix propagating inherit config in
SSHCluster
for non-bash shells (:pr:`3688`) Abdulelah Bin Mahfoodh - Add
Client.wait_to_workers
toClient
autosummary table (:pr:`3692`) James Bourbeau - Replace Bokeh Server with Tornado HTTPServer (:pr:`3658`) Matthew Rocklin
- Fix
dask-ssh
after removinglocal-directory
fromdask_scheduler
cli (:pr:`3684`) Abdulelah Bin Mahfoodh - Support preload modules in
Nanny
(:pr:`3678`) Matthew Rocklin - Refactor semaphore internals: make
_get_lease
synchronous (:pr:`3679`) Lucas Rademaker - Don't make task graphs too big (:pr:`3671`) Martin Durant
- Pass through
connection
/listen_args
as splatted keywords (:pr:`3674`) Matthew Rocklin - Run preload at import, start, and teardown (:pr:`3673`) Matthew Rocklin
- Use relative URL in scheduler dashboard (:pr:`3676`) Nicholas Smith
- Expose
Security
object as public API (:pr:`3675`) Matthew Rocklin - Add zoom tools to profile plots (:pr:`3672`) James Bourbeau
- Update
Scheduler.rebalance
return value when data is missing (:pr:`3670`) James Bourbeau
- Enable more UCX tests (:pr:`3667`) jakirkham
- Remove openssl 1.1.1d pin for Travis (:pr:`3668`) Jonathan J. Helmus
- More documentation for
Semaphore
(:pr:`3664`) Florian Jetter - Get CUDA context to finalize Numba
DeviceNDArray
(:pr:`3666`) jakirkham - Add Resources option to
get_task_stream
and calloutput_file
(:pr:`3653`) Prasun Anand - Add
Semaphore
extension (:pr:`3573`) Lucas Rademaker - Replace
ncores
withnthreads
in work stealing tests (:pr:`3615`) James Bourbeau - Clean up some test warnings (:pr:`3662`) Matthew Rocklin
- Write "why killed" docs (:pr:`3596`) Martin Durant
- Update Python version checking (:pr:`3660`) James Bourbeau
- Add newlines to ensure code formatting for
retire_workers
(:pr:`3661`) Rami Chowdhury - Clean up performance report test (:pr:`3655`) Matthew Rocklin
- Avoid diagnostics time in performance report (:pr:`3654`) Matthew Rocklin
- Introduce config for default task duration (:pr:`3642`) Gabriel Sailer
- UCX simplify receiving frames in
comm
(:pr:`3651`) jakirkham - Bump checkout GitHub action to v2 (:pr:`3649`) James Bourbeau
- Handle exception in
faulthandler
(:pr:`3646`) Jacob Tomlinson - Add prometheus metric for suspicious tasks (:pr:`3550`) Gabriel Sailer
- Remove
local-directory
keyword (:pr:`3620`) Prasun Anand - Don't create output Futures in Client when there are mixed Client Futures (:pr:`3643`) James Bourbeau
- Add link to
contributing.md
(:pr:`3621`) Prasun Anand - Update bokeh dependency in CI builds (:pr:`3637`) James Bourbeau
- UCX synchronize default stream only on CUDA frames (:pr:`3638`) Peter Andreas Entschev
- Add
as_completed.clear
method (:pr:`3617`) Matthew Rocklin - Drop unused line from
pack_frames_prelude
(:pr:`3634`) John Kirkham - Add logging message when closing idle dask scheduler (:pr:`3632`) Matthew Rocklin
- Include frame lengths of CUDA objects in
header["lengths"]
(:pr:`3631`) John Kirkham - Ensure
Client
connection pool semaphore attaches to theClient
event loop (:pr:`3546`) James Bourbeau - Remove dead stealing code (:pr:`3619`) Florian Jetter
- Check
nbytes
andtypes
before readingdata
(:pr:`3628`) John Kirkham - Ensure that we don't steal blacklisted fast tasks (:pr:`3591`) Florian Jetter
- Support async
Listener.stop
functions (:pr:`3613`) Matthew Rocklin - Add str/repr methods to
as_completed
(:pr:`3618`) Matthew Rocklin - Add backoff to comm connect attempts. (:pr:`3496`) Matthias Urlichs
- Make
Listeners
awaitable (:pr:`3611`) Matthew Rocklin - Increase number of visible mantissas in dashboard plots (:pr:`3585`) Scott Sievert
- Pin openssl to 1.1.1d for Travis (:pr:`3602`) Jacob Tomlinson
- Replace
tornado.queues
withasyncio.queues
(:pr:`3607`) James Bourbeau - Remove
dill
from CI environments (:pr:`3608`) Loïc Estève - Fix linting errors (:pr:`3604`) James Bourbeau
- Synchronize default CUDA stream before UCX send/recv (:pr:`3598`) Peter Andreas Entschev
- Add configuration for
Adaptive
arguments (:pr:`3509`) Gabriel Sailer - Change
Adaptive
docs to referenceadaptive_target
(:pr:`3597`) Julia Signell - Optionally compress on a frame-by-frame basis (:pr:`3586`) Matthew Rocklin
- Add Python version to version check (:pr:`3567`) James Bourbeau
- Import
tlz
(:pr:`3579`) John Kirkham - Pin
numpydoc
to avoid double escaped*
(:pr:`3530`) Gil Forsyth - Avoid
performance_report
crashing when a worker dies mid-compute (:pr:`3575`) Krishan Bhasin - Pin
bokeh
in CI builds (:pr:`3570`) James Bourbeau - Disable fast fail on GitHub Actions Windows CI (:pr:`3569`) James Bourbeau
- Fix typo in
Client.shutdown
docstring (:pr:`3562`) John Kirkham - Add
local_directory
option todask-ssh
(:pr:`3554`) Abdulelah Bin Mahfoodh
- Update
TaskGroup
remove logic (:pr:`3557`) James Bourbeau - Fix-up CuPy sparse serialization (:pr:`3556`) John Kirkham
- API docs for
LocalCluster
andSpecCluster
(:pr:`3548`) Tom Augspurger - Serialize sparse arrays (:pr:`3545`) John Kirkham
- Allow tasks with restrictions to be stolen (:pr:`3069`) Stan Seibert
- Use UCX default configuration instead of raising (:pr:`3544`) Peter Andreas Entschev
- Support using other serializers with
register_generic
(:pr:`3536`) John Kirkham - DOC: update to async await (:pr:`3543`) Tom Augspurger
- Use
pytest.raises
intest_ucx_config.py
(:pr:`3541`) John Kirkham - Fix/more ucx config options (:pr:`3539`) Benjamin Zaitlen
- Update heartbeat
CommClosedError
error handling (:pr:`3529`) James Bourbeau - Use
makedirs
when constructinglocal_directory
(:pr:`3538`) John Kirkham - Mark
None
as MessagePack serializable (:pr:`3537`) John Kirkham - Mark
bool
as MessagePack serializable (:pr:`3535`) John Kirkham - Use 'temporary-directory' from
dask.config
for Nanny's directory (:pr:`3531`) John Kirkham - Add try-except around getting source code in performance report (:pr:`3505`) Matthew Rocklin
- Fix typo in docstring (:pr:`3528`) Davis Bennett
- Make work stealing callback time configurable (:pr:`3523`) Lucas Rademaker
- RMM/UCX Config Flags (:pr:`3515`) Benjamin Zaitlen
- Revise develop-docs: conda env example (:pr:`3406`) Darren Weber
- Remove
import ucp
from the top ofucx.py
(:pr:`3510`) Peter Andreas Entschev - Rename
logs
toget_logs
(:pr:`3473`) Jacob Tomlinson - Stop keep alives when worker reconnecting to the scheduler (:pr:`3493`) Jacob Tomlinson
- Add dask serialization of CUDA objects (:pr:`3482`) John Kirkham
- Suppress cuML
ImportError
(:pr:`3499`) John Kirkham - Msgpack 1.0 compatibility (:pr:`3494`) James Bourbeau
- Register cuML serializers (:pr:`3485`) John Kirkham
- Check exact equality for worker state (:pr:`3483`) Brett Naul
- Serialize 1-D, contiguous,
uint8
CUDA frames (:pr:`3475`) John Kirkham - Update NumPy array serialization to handle non-contiguous slices (:pr:`3474`) James Bourbeau
- Propose fix for collection based resources docs (:pr:`3480`) Chris Roat
- Remove
--verbose
flag from CI runs (:pr:`3484`) Matthew Rocklin - Do not duplicate messages in scheduler report (:pr:`3477`) Jakub Beránek
- Register Dask cuDF serializers (:pr:`3478`) John Kirkham
- Add support for Python 3.8 (:pr:`3249`) James Bourbeau
- Add last seen column to worker table and highlight errant workers (:pr:`3468`) kaelgreco
- Change default value of
local_directory
from empty string toNone
(:pr:`3441`) condoratberlin - Clear old docs (:pr:`3458`) Matthew Rocklin
- Change default multiprocessing behavior to spawn (:pr:`3461`) Matthew Rocklin
- Split dashboard host on additional slashes to handle inproc (:pr:`3466`) Jacob Tomlinson
- Update
locality.rst
(:pr:`3470`) Dustin Tindall - Minor
gen.Return
cleanup (:pr:`3469`) James Bourbeau - Update comparison logic for worker state (:pr:`3321`) rockwellw
- Update minimum
tblib
version to 1.6.0 (:pr:`3451`) James Bourbeau - Add total row to workers plot in dashboard (:pr:`3464`) Julia Signell
- Workaround
RecursionError
on profile data (:pr:`3455`) Tom Augspurger - Include code and summary in performance report (:pr:`3462`) Matthew Rocklin
- Skip
test_open_close_many_workers
on Python 3.6 (:pr:`3459`) Matthew Rocklin - Support serializing/deserializing
rmm.DeviceBuffer
s (:pr:`3442`) John Kirkham - Always add new
TaskGroup
toTaskPrefix
(:pr:`3322`) James Bourbeau - Rerun
black
on the code base (:pr:`3444`) John Kirkham - Ensure
__causes__
s of exceptions raised on workers are serialized (:pr:`3430`) Alex Adamson - Adjust
numba.cuda
import and add check (:pr:`3446`) John Kirkham - Fix name of Numba serialization test (:pr:`3447`) John Kirkham
- Checks for command parameters in
ssh2
(:pr:`3078`) Peter Andreas Entschev - Update
worker_kwargs
description inLocalCluster
constructor (:pr:`3438`) James Bourbeau - Ensure scheduler updates task and worker states after successful worker data deletion (:pr:`3401`) James Bourbeau
- Avoid
loop=
keyword in asyncio coordination primitives (:pr:`3437`) Matthew Rocklin - Call pip as a module to avoid warnings (:pr:`3436`) Cyril Shcherbin
- Add documentation of parameters in coordination primitives (:pr:`3434`) Søren Fuglede Jørgensen
- Replace
tornado.locks
with asyncio for Events/Locks/Conditions/Semaphore (:pr:`3397`) Matthew Rocklin - Remove object from class hierarchy (:pr:`3432`) Anderson Banihirwe
- Add
dashboard_link
property toClient
(:pr:`3429`) Jacob Tomlinson - Allow memory monitor to evict data more aggressively (:pr:`3424`) fjetter
- Make
_get_ip
return an IP address when defaulting (:pr:`3418`) Pierre Glaser - Support version checking with older versions of Dask (:pr:`3390`) Igor Gotlibovych
- Add Mac OS build to CI (:pr:`3358`) James Bourbeau
- Fixed
ZeroDivisionError
in dashboard when no workers were present (:pr:`3407`) James Bourbeau - Respect the
dashboard-prefix
when redirecting from the root (:pr:`3387`) Chrysostomos Nanakos - Allow enabling / disabling work-stealing after the cluster has started (:pr:`3410`) John Kirkham
- Support
*args
and**kwargs
in offload (:pr:`3392`) Matthew Rocklin - Add lifecycle hooks to SchedulerPlugin (:pr:`3391`) Matthew Rocklin
- Raise
RuntimeError
if no running loop (:pr:`3385`) James Bourbeau - Fix
get_running_loop
import (:pr:`3383`) James Bourbeau - Get JavaScript document location instead of window and handle proxied url (:pr:`3382`) Jacob Tomlinson
- Move Windows CI to GitHub Actions (:pr:`3373`) Jacob Tomlinson
- Add client join and leave hooks (:pr:`3371`) Jacob Tomlinson
- Add cluster map dashboard (:pr:`3361`) Jacob Tomlinson
- Close connection comm on retry (:pr:`3365`) James Bourbeau
- Fix scheduler state in case of worker name collision (:pr:`3366`) byjott
- Add
--worker-class
option todask-worker
CLI (:pr:`3364`) James Bourbeau - Remove
locale
check that fails on OS X (:pr:`3360`) Jacob Tomlinson - Rework version checking (:pr:`2627`) Matthew Rocklin
- Add websocket scheduler plugin (:pr:`3335`) Jacob Tomlinson
- Return task in
dask-worker
on_signal
function (:pr:`3354`) James Bourbeau - Fix failures on mixed integer/string worker names (:pr:`3352`) Benedikt Reinartz
- Avoid calling
nbytes
multiple times when sending data (:pr:`3349`) Markus Mohrhard - Avoid setting event loop policy if within IPython kernel and no running event loop (:pr:`3336`) Mana Borwornpadungkitti
- Relax intermittent failing
test_profile_server
(:pr:`3346`) Matthew Rocklin
- Add lock around dumps_function cache (:pr:`3337`) Matthew Rocklin
- Add setuptools to dependencies (:pr:`3320`) James Bourbeau
- Use TaskPrefix.name in Graph layout (:pr:`3328`) Matthew Rocklin
- Add missing " in performance report example (:pr:`3329`) John Kirkham
- Add performance report docs and color definitions to docs (:pr:`3325`) Benjamin Zaitlen
- Switch startstops to dicts and add worker name to transfer (:pr:`3319`) Jacob Tomlinson
- Add plugin entry point for out-of-tree comms library (:pr:`3305`) Patrick Sodré
- All scheduler task states in prometheus (:pr:`3307`) fjetter
- Use worker name in logs (:pr:`3309`) Stephan Erb
- Add TaskGroup and TaskPrefix scheduler state (:pr:`3262`) Matthew Rocklin
- Update latencies with heartbeats (:pr:`3310`) fjetter
- Update inlining Futures in task graph in Client._graph_to_futures (:pr:`3303`) James Bourbeau
- Use hostname as default IP address rather than localhost (:pr:`3308`) Matthew Rocklin
- Clean up flaky test_nanny_throttle (:pr:`3295`) Tom Augspurger
- Add lock to scheduler for sensitive operations (:pr:`3259`) Matthew Rocklin
- Log address for each of the Scheduler listerners (:pr:`3306`) Matthew Rocklin
- Make ConnectionPool.close asynchronous (:pr:`3304`) Matthew Rocklin
- Add
dask-spec
CLI tool (:pr:`3090`) Matthew Rocklin - Connectionpool: don't hand out closed connections (:pr:`3301`) byjott
- Retry operations on network issues (:pr:`3294`) byjott
- Skip
Security.temporary()
tests if cryptography not installed (:pr:`3302`) James Bourbeau - Support multiple listeners in the scheduler (:pr:`3288`) Matthew Rocklin
- Updates RMM comment to the correct release (:pr:`3299`) John Kirkham
- Add title to
performance_report
(:pr:`3298`) Matthew Rocklin - Forgot to fix slow test (:pr:`3297`) Benjamin Zaitlen
- Update
SSHCluster
docstring parameters (:pr:`3296`) James Bourbeau worker.close()
awaitsbatched_stream.close()
(:pr:`3291`) Mads R. B. Kristensen- Fix asynchronous listener in UCX (:pr:`3292`) Benjamin Zaitlen
- Avoid repeatedly adding deps to already in memory stack (:pr:`3293`) James Bourbeau
- xfail ucx empty object typed dataframe (:pr:`3279`) Benjamin Zaitlen
- Fix
distributed.wait
documentation (:pr:`3289`) Tom Rochette - Move Python 3 syntax tests into main tests (:pr:`3281`) Matthew Rocklin
- xfail
test_workspace_concurrency
for Python 3.6 (:pr:`3283`) Matthew Rocklin - Add
performance_report
context manager for static report generation (:pr:`3282`) Matthew Rocklin - Update function serialization caches with custom LRU class (:pr:`3260`) James Bourbeau
- Make
Listener.start
asynchronous (:pr:`3278`) Matthew Rocklin - Remove
dask-submit
anddask-remote
(:pr:`3280`) Matthew Rocklin - Worker profile server (:pr:`3274`) Matthew Rocklin
- Improve bandwidth workers plot (:pr:`3273`) Matthew Rocklin
- Make profile coroutines consistent between
Scheduler
andWorker
(:pr:`3277`) Matthew Rocklin - Enable saving profile information from server threads (:pr:`3271`) Matthew Rocklin
- Remove memory use plot (:pr:`3269`) Matthew Rocklin
- Add offload size to configuration (:pr:`3270`) Matthew Rocklin
- Fix layout scaling on profile plots (:pr:`3268`) Jacob Tomlinson
- Set
x_range
in CPU plot based on the number of threads (:pr:`3266`) Matthew Rocklin - Use base-2 values for byte-valued axes in dashboard (:pr:`3267`) Matthew Rocklin
- Robust gather in case of connection failures (:pr:`3246`) fjetter
- Use
DeviceBuffer
from newer RMM releases (:pr:`3261`) John Kirkham - Fix dev requirements for pytest (:pr:`3264`) Elliott Sales de Andrade
- Add validate options to configuration (:pr:`3258`) Matthew Rocklin
- Fix hanging worker when the scheduler leaves (:pr:`3250`) Tom Augspurger
- Fix NumPy writeable serialization bug (:pr:`3253`) James Bourbeau
- Skip
numba.cuda
tests if CUDA is not available (:pr:`3255`) Peter Andreas Entschev - Add new dashboard plot for memory use by key (:pr:`3243`) Matthew Rocklin
- Fix
array.shape()
->array.shape
(:pr:`3247`) Jed Brown - Fixed typos in
pubsub.py
(:pr:`3244`) He Jia - Fixed cupy array going out of scope (:pr:`3240`) Mads R. B. Kristensen
- Remove
gen.coroutine
usage in scheduler (:pr:`3242`) Jim Crist-Harif - Use
inspect.isawaitable
where relevant (:pr:`3241`) Jim Crist-Harif
- Add UCX config values (:pr:`3135`) Matthew Rocklin
- Relax test_MultiWorker (:pr:`3210`) Matthew Rocklin
- Avoid ucp.init at import time (:pr:`3211`) Matthew Rocklin
- Clean up rpc to avoid intermittent test failure (:pr:`3215`) Matthew Rocklin
- Respect protocol if given to Scheduler (:pr:`3212`) Matthew Rocklin
- Use legend_field= keyword in bokeh plots (:pr:`3218`) Matthew Rocklin
- Cache psutil.Process object in Nanny (:pr:`3207`) Matthew Rocklin
- Replace gen.sleep with asyncio.sleep (:pr:`3208`) Matthew Rocklin
- Avoid offloading serialization for small messages (:pr:`3224`) Matthew Rocklin
- Add desired_workers metric (:pr:`3221`) Gabriel Sailer
- Fail fast when importing distributed.comm.ucx (:pr:`3228`) Matthew Rocklin
- Add module name to Future repr (:pr:`3231`) Matthew Rocklin
- Add name to Pub/Sub repr (:pr:`3235`) Matthew Rocklin
- Import CPU_COUNT from dask.system (:pr:`3199`) James Bourbeau
- Efficiently serialize zero strided NumPy arrays (:pr:`3180`) James Bourbeau
- Cache function deserialization in workers (:pr:`3234`) Matthew Rocklin
- Respect ordering of futures in futures_of (:pr:`3236`) Matthew Rocklin
- Bump dask dependency to 2.7.0 (:pr:`3237`) James Bourbeau
- Avoid setting inf x_range (:pr:`3229`) rockwellw
- Clear task stream based on recent behavior (:pr:`3200`) Matthew Rocklin
- Use the percentage field for profile plots (:pr:`3238`) Matthew Rocklin
This release drops support for Python 3.5
- Adds badges to README.rst [skip ci] (:pr:`3152`) James Bourbeau
- Don't overwrite self.address if it is present (:pr:`3153`) Gil Forsyth
- Remove outdated references to debug scheduler and worker bokeh pages. (:pr:`3160`) darindf
- Update CONTRIBUTING.md (:pr:`3159`) Jacob Tomlinson
- Add Prometheus metric for a worker's executing tasks count (:pr:`3163`) darindf
- Update Prometheus documentation (:pr:`3165`) darindf
- Fix Numba serialization when strides is None (:pr:`3166`) Peter Andreas Entschev
- Await cluster in Adaptive.recommendations (:pr:`3168`) Simon Boothroyd
- Support automatic TLS (:pr:`3164`) Jim Crist
- Avoid swamping high-memory workers with data requests (:pr:`3071`) Tom Augspurger
- Update UCX variables to use sockcm by default (:pr:`3177`) Peter Andreas Entschev
- Get protocol in Nanny/Worker from scheduler address (:pr:`3175`) Peter Andreas Entschev
- Add worker and tasks state for Prometheus data collection (:pr:`3174`) darindf
- Use async def functions for offload to/from_frames (:pr:`3171`) Mads R. B. Kristensen
- Subprocesses inherit the global dask config (:pr:`3192`) Mads R. B. Kristensen
- XFail test_open_close_many_workers (:pr:`3194`) Matthew Rocklin
- Drop Python 3.5 (:pr:`3179`) James Bourbeau
- UCX: avoid double init after fork (:pr:`3178`) Mads R. B. Kristensen
- Silence warning when importing while offline (:pr:`3203`) James A. Bednar
- Adds docs to Client methods for resources, actors, and traverse (:pr:`2851`) IPetrik
- Add test for concurrent scatter operations (:pr:`2244`) Matthew Rocklin
- Expand async docs (:pr:`2293`) Dave Hirschfeld
- Add PatchedDeviceArray to drop stride attribute for cupy<7.0 (:pr:`3198`) Richard J Zamora
- Refactor dashboard module (:pr:`3138`) Jacob Tomlinson
- Use
setuptools.find_packages
insetup.py
(:pr:`3150`) Matthew Rocklin - Move death timeout logic up to
Node.start
(:pr:`3115`) Matthew Rocklin - Only include metric in
WorkerTable
if it is a scalar (:pr:`3140`) Matthew Rocklin - Add
Nanny(config={...})
keyword (:pr:`3134`) Matthew Rocklin - Xfail
test_worksapce_concurrency
on Python 3.6 (:pr:`3132`) Matthew Rocklin - Extend Worker plugin API with transition method (:pr:`2994`) matthieubulte
- Raise exception if the user passes in unused keywords to
Client
(:pr:`3117`) Jonathan De Troye - Move new
SSHCluster
to top level (:pr:`3128`) Matthew Rocklin - Bump dask dependency (:pr:`3124`) Jim Crist
- Make dask-worker close quietly when given sigint signal (:pr:`3116`) Matthew Rocklin
- Replace use of tornado.gen with asyncio in dask-worker (:pr:`3114`) Matthew Rocklin
- UCX: allocate CUDA arrays using RMM and Numba (:pr:`3109`) Mads R. B. Kristensen
- Support calling cluster.scale as async method (:pr:`3110`) Jim Crist
- Identify lost workers in SpecCluster based on address not name (:pr:`3088`) James Bourbeau
- Add Client.shutdown method (:pr:`3106`) Matthew Rocklin
- Collect worker-worker and type bandwidth information (:pr:`3094`) Matthew Rocklin
- Send noise over the wire to keep dask-ssh connection alive (:pr:`3105`) Gil Forsyth
- Retry scheduler connect multiple times (:pr:`3104`) Jacob Tomlinson
- Add favicon of logo to the dashboard (:pr:`3095`) James Bourbeau
- Remove utils.py functions for their dask/utils.py equivalents (:pr:`3042`) Matthew Rocklin
- Lower default bokeh log level (:pr:`3087`) Philipp Rudiger
- Check if self.cluster.scheduler is a local scheduler (:pr:`3099`) Jacob Tomlinson
- Support clusters that don't have .security or ._close methods (:pr:`3100`) Matthew Rocklin
- Use the new UCX Python bindings (:pr:`3059`) Mads R. B. Kristensen
- Fix worker preload config (:pr:`3027`) byjott
- Fix widget with spec that generates multiple workers (:pr:`3067`) Loïc Estève
- Make Client.get_versions async friendly (:pr:`3064`) Jacob Tomlinson
- Add configuration option for longer error tracebacks (:pr:`3086`) Daniel Farrell
- Have Client get Security from passed Cluster (:pr:`3079`) Matthew Rocklin
- Respect Cluster.dashboard_link in Client._repr_html_ if it exists (:pr:`3077`) Matthew Rocklin
- Add monitoring with dask cluster docs (:pr:`3072`) Arpit Solanki
- Protocol of cupy and numba handles serialization exclusively (:pr:`3047`) Mads R. B. Kristensen
- Allow specification of worker type in SSHCLuster (:pr:`3061`) Jacob Tomlinson
- Use Cluster.scheduler_info for workers= value in repr (:pr:`3058`) Matthew Rocklin
- Allow SpecCluster to scale by memory and cores (:pr:`3057`) Matthew Rocklin
- Allow full script in preload inputs (:pr:`3052`) Matthew Rocklin
- Check multiple cgroups dirs, ceil fractional cpus (:pr:`3056`) Jim Crist
- Add blurb about disabling work stealing (:pr:`3055`) Chris White
- Remove six (:pr:`3045`) Matthew Rocklin
- Add missing test data to sdist tarball (:pr:`3050`) Elliott Sales de Andrade
- Use mock from unittest standard library (:pr:`3049`) Elliott Sales de Andrade
- Use cgroups resource limits to determine default threads and memory (:pr:`3039`) Jim Crist
- Move task deserialization to immediately before task execution (:pr:`3015`) James Bourbeau
- Drop joblib shim module in distributed (:pr:`3040`) John Kirkham
- Redirect configuration doc page (:pr:`3038`) Matthew Rocklin
- Support
--name 0
and--nprocs
keywords in dask-worker cli (:pr:`3037`) Matthew Rocklin - Remove lost workers from
SpecCluster.workers
(:pr:`2990`) Guillaume Eynard-Bontemps - Clean up
test_local.py::test_defaults
(:pr:`3017`) Matthew Rocklin - Replace print statement in
Queue.__init__
with debug message (:pr:`3035`) Mikhail Akimov - Set the
x_range
limit of the Meory utilization plot to memory-limit (:pr:`3034`) Matthew Rocklin - Rely on cudf codebase for cudf serialization (:pr:`2998`) Benjamin Zaitlen
- Add fallback html repr for Cluster (:pr:`3023`) Jim Crist
- Add support for zstandard compression to comms (:pr:`2970`) Abael He
- Avoid collision when using
os.environ
indashboard_link
(:pr:`3021`) Matthew Rocklin - Fix
ConnectionPool
limit handling (:pr:`3005`) byjott - Support Spec jobs that generate multiple workers (:pr:`3013`) Matthew Rocklin
- Tweak
Logs
styling (:pr:`3012`) Jim Crist - Better name for cudf deserialization function name (:pr:`3008`) Benjamin Zaitlen
- Make
spec.ProcessInterface
a valid no-op worker (:pr:`3004`) Matthew Rocklin - Return dictionaries from
new_worker_spec
rather than name/worker pairs (:pr:`3000`) Matthew Rocklin - Fix minor typo in documentation (:pr:`3002`) Mohammad Noor
- Permit more keyword options when scaling with cores and memory (:pr:`2997`) Matthew Rocklin
- Add
cuda_ipc
to UCX environment for NVLink (:pr:`2996`) Benjamin Zaitlen - Add
threads=
andmemory=
to Cluster and Client reprs (:pr:`2995`) Matthew Rocklin - Fix PyNVML initialization (:pr:`2993`) Richard J Zamora
- Skip exceptions in startup information (:pr:`2991`) Jacob Tomlinson
- Add support for separate external address for SpecCluster scheduler (:pr:`2963`) Jacob Tomlinson
- Defer cudf serialization/deserialization to that library (:pr:`2881`) Benjamin Zaitlen
- Workaround for hanging test now calls ucp.fin() (:pr:`2967`) Mads R. B. Kristensen
- Remove unnecessary bullet point (:pr:`2972`) Pav A
- Directly import progress from diagnostics.progressbar (:pr:`2975`) Matthew Rocklin
- Handle buffer protocol objects in ensure_bytes (:pr:`2969`) Tom Augspurger
- Fix documentatation syntax and tree (:pr:`2981`) Pav A
- Improve get_ip_interface error message when interface does not exist (:pr:`2964`) Loïc Estève
- Add cores= and memory= keywords to scale (:pr:`2974`) Matthew Rocklin
- Make workers robust to bad custom metrics (:pr:`2984`) Matthew Rocklin
- Except all exceptions when checking
pynvml
(:pr:`2961`) Matthew Rocklin - Pass serialization down through small base collections (:pr:`2948`) Peter Andreas Entschev
- Use
pytest.warning(Warning)
rather thanException
(:pr:`2958`) Matthew Rocklin - Allow
server_kwargs
to override defaults in dashboard (:pr:`2955`) Bruce Merry - Update
utils_perf.py
(:pr:`2954`) Shayan Amani - Normalize names with
str
inretire_workers
(:pr:`2949`) Matthew Rocklin - Update
client.py
(:pr:`2951`) Shayan Amani - Add
GPUCurrentLoad
dashboard plots (:pr:`2944`) Matthew Rocklin - Pass GPU diagnostics from worker to scheduler (:pr:`2932`) Matthew Rocklin
- Import from
collections.abc
(:pr:`2938`) Jim Crist - Fixes Worker docstring formatting (:pr:`2939`) James Bourbeau
- Redirect setup docs to docs.dask.org (:pr:`2936`) Matthew Rocklin
- Wrap offload in
gen.coroutine
(:pr:`2934`) Matthew Rocklin - Change
TCP.close
to a coroutine to avoid task pending warning (:pr:`2930`) Matthew Rocklin - Fixup black string normalization (:pr:`2929`) Jim Crist
- Move core functionality from
SpecCluster
toCluster
(:pr:`2913`) Matthew Rocklin - Add aenter/aexit protocols to
ProcessInterface
(:pr:`2927`) Matthew Rocklin - Add real-time CPU utilization plot to dashboard (:pr:`2922`) Matthew Rocklin
- Always kill processes in clean tests, even if we don't check (:pr:`2924`) Matthew Rocklin
- Add timeouts to processes in SSH tests (:pr:`2925`) Matthew Rocklin
- Add documentation around
spec.ProcessInterface
(:pr:`2923`) Matthew Rocklin - Cleanup async warnings in tests (:pr:`2920`) Matthew Rocklin
- Give 404 when requesting nonexistent tasks or workers (:pr:`2921`) Martin Durant
- Raise informative warning when rescheduling an unknown task (:pr:`2916`) James Bourbeau
- Fix docstring (:pr:`2917`) Martin Durant
- Add keep-alive message between worker and scheduler (:pr:`2907`) Matthew Rocklin
- Rewrite
Adaptive
/SpecCluster
to support slowly arriving workers (:pr:`2904`) Matthew Rocklin - Call heartbeat rather than reconnect on disconnection (:pr:`2906`) Matthew Rocklin
- Respect security configuration in LocalCluster (:pr:`2822`) Russ Bubley
- Add Nanny to worker docs (:pr:`2826`) Christian Hudon
- Don't make False add-keys report to scheduler (:pr:`2421`) tjb900
- Include type name in SpecCluster repr (:pr:`2834`) Jacob Tomlinson
- Extend prometheus metrics endpoint (:pr:`2833`) Gabriel Sailer
- Add alternative SSHCluster implementation (:pr:`2827`) Matthew Rocklin
- Dont reuse closed worker in get_worker (:pr:`2841`) Pierre Glaser
- SpecCluster: move init logic into start (:pr:`2850`) Jacob Tomlinson
- Document distributed.Reschedule in API docs (:pr:`2860`) James Bourbeau
- Add fsspec to installation of test builds (:pr:`2859`) Martin Durant
- Make await/start more consistent across Scheduler/Worker/Nanny (:pr:`2831`) Matthew Rocklin
- Add cleanup fixture for asyncio tests (:pr:`2866`) Matthew Rocklin
- Use only remote connection to scheduler in Adaptive (:pr:`2865`) Matthew Rocklin
- Add Server.finished async function (:pr:`2864`) Matthew Rocklin
- Align text and remove bullets in Client HTML repr (:pr:`2867`) Matthew Rocklin
- Test dask-scheduler --idle-timeout flag (:pr:`2862`) Matthew Rocklin
- Remove
Client.upload_environment
(:pr:`2877`) Jim Crist - Replace gen.coroutine with async/await in core (:pr:`2871`) Matthew Rocklin
- Forcefully kill all processes before each test (:pr:`2882`) Matthew Rocklin
- Cleanup Security class and configuration (:pr:`2873`) Jim Crist
- Remove unused variable in SpecCluster scale down (:pr:`2870`) Jacob Tomlinson
- Add SpecCluster ProcessInterface (:pr:`2874`) Jacob Tomlinson
- Add Log(str) and Logs(dict) classes for nice HTML reprs (:pr:`2875`) Jacob Tomlinson
- Pass Client._asynchronous to Cluster._asynchronous (:pr:`2890`) Matthew Rocklin
- Add default logs method to Spec Cluster (:pr:`2889`) Matthew Rocklin
- Add processes keyword back into clean (:pr:`2891`) Matthew Rocklin
- Update black (:pr:`2901`) Matthew Rocklin
- Move Worker.local_dir attribute to Worker.local_directory (:pr:`2900`) Matthew Rocklin
- Link from TapTools to worker info pages in dashboard (:pr:`2894`) Matthew Rocklin
- Avoid exception in Client._ensure_connected if closed (:pr:`2893`) Matthew Rocklin
- Convert Pythonic kwargs to CLI Keywords for SSHCluster (:pr:`2898`) Matthew Rocklin
- Use kwargs in CLI (:pr:`2899`) Matthew Rocklin
- Name SSHClusters by providing name= keyword to SpecCluster (:pr:`2903`) Matthew Rocklin
- Request feed of worker information from Scheduler to SpecCluster (:pr:`2902`) Matthew Rocklin
- Clear out compatibillity file (:pr:`2896`) Matthew Rocklin
- Remove future imports (:pr:`2897`) Matthew Rocklin
- Use click's show_default=True in relevant places (:pr:`2838`) Christian Hudon
- Close workers more gracefully (:pr:`2905`) Matthew Rocklin
- Close workers gracefully with --lifetime keywords (:pr:`2892`) Matthew Rocklin
- Add closing <li> tags to Client._repr_html_ (:pr:`2911`) Matthew Rocklin
- Add endline spacing in Logs._repr_html_ (:pr:`2912`) Matthew Rocklin
- Fix typo that prevented error message (:pr:`2825`) Russ Bubley
- Remove
dask-mpi
(:pr:`2824`) Matthew Rocklin - Updates to use
update_graph
in task journey docs (:pr:`2821`) James Bourbeau - Fix Client repr with
memory_info=None
(:pr:`2816`) Matthew Rocklin - Fix case where key, rather than
TaskState
, could end up ints.waiting_on
(:pr:`2819`) tjb900 - Use Keyword-only arguments (:pr:`2814`) Matthew Rocklin
- Relax check for worker references in cluster context manager (:pr:`2813`) Matthew Rocklin
- Add HTTPS support for the dashboard (:pr:`2812`) Jim Crist
- Use
dask.utils.format_bytes
(:pr:`2810`) Tom Augspurger
We neglected to include python_requires=
in our setup.py file, resulting in
confusion for Python 2 users who erroneously get packages for 2.0.0.
This is fixed in 2.0.1 and we have removed the 2.0.0 files from PyPI.
- Add python_requires entry to setup.py (:pr:`2807`) Matthew Rocklin
- Correctly manage tasks beyond deque limit in TaskStream plot (:pr:`2797`) Matthew Rocklin
- Fix diagnostics page for memory_limit=None (:pr:`2770`) Brett Naul
- Drop support for Python 2
- Relax warnings before release (:pr:`2796`) Matthew Rocklin
- Deprecate --bokeh/--no-bokeh CLI (:pr:`2800`) Tom Augspurger
- Typo in bokeh service_kwargs for dask-worker (:pr:`2783`) Tom Augspurger
- Update command line cli options docs (:pr:`2794`) James Bourbeau
- Remove "experimental" from TLS docs (:pr:`2793`) James Bourbeau
- Add warnings around ncores= keywords (:pr:`2791`) Matthew Rocklin
- Add --version option to scheduler and worker CLI (:pr:`2782`) Tom Augspurger
- Raise when workers initialization times out (:pr:`2784`) Tom Augspurger
- Replace ncores with nthreads throughout codebase (:pr:`2758`) Matthew Rocklin
- Add unknown pytest markers (:pr:`2764`) Tom Augspurger
- Delay lookup of allowed failures. (:pr:`2761`) Tom Augspurger
- Change address -> worker in ColumnDataSource for nbytes plot (:pr:`2755`) Matthew Rocklin
- Remove module state in Prometheus Handlers (:pr:`2760`) Matthew Rocklin
- Add stress test for UCX (:pr:`2759`) Matthew Rocklin
- Add nanny logs (:pr:`2744`) Tom Augspurger
- Move some of the adaptive logic into the scheduler (:pr:`2735`) Matthew Rocklin
- Add SpecCluster.new_worker_spec method (:pr:`2751`) Matthew Rocklin
- Worker dashboard fixes (:pr:`2747`) Matthew Rocklin
- Add async context managers to scheduler/worker classes (:pr:`2745`) Matthew Rocklin
- Fix the resource key representation before sending graphs (:pr:`2733`) Michael Spiegel
- Allow user to configure whether workers are daemon. (:pr:`2739`) Caleb
- Pin pytest >=4 with pip in appveyor and python 3.5 (:pr:`2737`) Matthew Rocklin
- Add Experimental UCX Comm (:pr:`2591`) Ben Zaitlen Tom Augspurger Matthew Rocklin
- Close nannies gracefully (:pr:`2731`) Matthew Rocklin
- add kwargs to progressbars (:pr:`2638`) Manuel Garrido
- Add back LocalCluster.__repr__. (:pr:`2732`) Loïc Estève
- Move bokeh module to dashboard (:pr:`2724`) Matthew Rocklin
- Close clusters at exit (:pr:`2730`) Matthew Rocklin
- Add SchedulerPlugin TaskState example (:pr:`2622`) Matt Nicolls
- Add SpecificationCluster (:pr:`2675`) Matthew Rocklin
- Replace register_worker_callbacks with worker plugins (:pr:`2453`) Matthew Rocklin
- Proxy worker dashboards from scheduler dashboard (:pr:`2715`) Ben Zaitlen
- Add docstring to Scheduler.check_idle_saturated (:pr:`2721`) Matthew Rocklin
- Refer to LocalCluster in Client docstring (:pr:`2719`) Matthew Rocklin
- Remove special casing of Scikit-Learn BaseEstimator serialization (:pr:`2713`) Matthew Rocklin
- Fix two typos in Pub class docstring (:pr:`2714`) Magnus Nord
- Support uploading files with multiple modules (:pr:`2587`) Sam Grayson
- Change the main workers bokeh page to /status (:pr:`2689`) Ben Zaitlen
- Cleanly stop periodic callbacks in Client (:pr:`2705`) Matthew Rocklin
- Disable pan tool for the Progress, Byte Stored and Tasks Processing plot (:pr:`2703`) Mathieu Dugré
- Except errors in Nanny's memory monitor if process no longer exists (:pr:`2701`) Matthew Rocklin
- Handle heartbeat when worker has just left (:pr:`2702`) Matthew Rocklin
- Modify styling of histograms for many-worker dashboard plots (:pr:`2695`) Mathieu Dugré
- Add method to wait for n workers before continuing (:pr:`2688`) Daniel Farrell
- Support computation on delayed(None) (:pr:`2697`) Matthew Rocklin
- Cleanup localcluster (:pr:`2693`) Matthew Rocklin
- Use 'temporary-directory' from dask.config for Worker's directory (:pr:`2654`) Matthew Rocklin
- Remove support for Iterators and Queues (:pr:`2671`) Matthew Rocklin
This is a small bugfix release due to a config change upstream.
- Use config accessor method for "scheduler-address" (:pr:`2676`) James Bourbeau
- Add Type Attribute to TaskState (:pr:`2657`) Matthew Rocklin
- Add waiting task count to progress title bar (:pr:`2663`) James Bourbeau
- DOC: Clean up reference to cluster object (:pr:`2664`) K.-Michael Aye
- Allow scheduler to politely close workers as part of shutdown (:pr:`2651`) Matthew Rocklin
- Check direct_to_workers before using get_worker in Client (:pr:`2656`) Matthew Rocklin
- Fixed comment regarding keeping existing level if less verbose (:pr:`2655`) Brett Randall
- Add idle timeout to scheduler (:pr:`2652`) Matthew Rocklin
- Avoid deprecation warnings (:pr:`2653`) Matthew Rocklin
- Use an LRU cache for deserialized functions (:pr:`2623`) Matthew Rocklin
- Rename Worker._close to Worker.close (:pr:`2650`) Matthew Rocklin
- Add Comm closed bookkeeping (:pr:`2648`) Matthew Rocklin
- Explain LocalCluster behavior in Client docstring (:pr:`2647`) Matthew Rocklin
- Add last worker into KilledWorker exception to help debug (:pr:`2610`) @plbertrand
- Set working worker class for dask-ssh (:pr:`2646`) Martin Durant
- Add as_completed methods to docs (:pr:`2642`) Jim Crist
- Add timeout to Client._reconnect (:pr:`2639`) Jim Crist
- Limit test_spill_by_default memory, reenable it (:pr:`2633`) Peter Andreas Entschev
- Use proper address in worker -> nanny comms (:pr:`2640`) Jim Crist
- Fix deserialization of bytes chunks larger than 64MB (:pr:`2637`) Peter Andreas Entschev
- Adaptive: recommend close workers when any are idle (:pr:`2330`) Michael Delgado
- Increase GC thresholds (:pr:`2624`) Matthew Rocklin
- Add interface= keyword to LocalCluster (:pr:`2629`) Matthew Rocklin
- Add worker_class argument to LocalCluster (:pr:`2625`) Matthew Rocklin
- Remove Python 2.7 from testing matrix (:pr:`2631`) Matthew Rocklin
- Add number of trials to diskutils test (:pr:`2630`) Matthew Rocklin
- Fix parameter name in LocalCluster docstring (:pr:`2626`) Loïc Estève
- Integrate stacktrace for low-level profiling (:pr:`2575`) Peter Andreas Entschev
- Apply Black to standardize code styling (:pr:`2614`) Matthew Rocklin
- added missing whitespace to start_worker cmd (:pr:`2613`) condoratberlin
- Updated logging module doc links from docs.python.org/2 to docs.python.org/3. (:pr:`2635`) Brett Randall
- Add basic health endpoints to scheduler and worker bokeh. (:pr:`2607`) amerkel2
- Improved description accuracy of --memory-limit option. (:pr:`2601`) Brett Randall
- Check self.dependencies when looking at dependent tasks in memory (:pr:`2606`) deepthirajagopalan7
- Add RabbitMQ SchedulerPlugin example (:pr:`2604`) Matt Nicolls
- add resources to scheduler update_graph plugin (:pr:`2603`) Matt Nicolls
- Use ensure_bytes in serialize_error (:pr:`2588`) Matthew Rocklin
- Specify data storage explicitly from Worker constructor (:pr:`2600`) Matthew Rocklin
- Change bokeh port keywords to dashboard_address (:pr:`2589`) Matthew Rocklin
- .detach_() pytorch tensor to serialize data as numpy array. (:pr:`2586`) Muammar El Khatib
- Add warning if creating scratch directories takes a long time (:pr:`2561`) Matthew Rocklin
- Fix typo in pub-sub doc. (:pr:`2599`) Loïc Estève
- Allow return_when='FIRST_COMPLETED' in wait (:pr:`2598`) Nikos Tsaousis
- Forward kwargs through Nanny to Worker (:pr:`2596`) Brian Chu
- Use ensure_dict instead of dict (:pr:`2594`) James Bourbeau
- Specify protocol in LocalCluster (:pr:`2489`) Matthew Rocklin
- Fix LocalCluster to not overallocate memory when overcommitting threads per worker (:pr:`2541`) George Sakkis
- Make closing resilient to lacking an address (:pr:`2542`) Matthew Rocklin
- fix typo in comment (:pr:`2546`) Brett Jurman
- Fix double init of prometheus metrics (:pr:`2544`) Marco Neumann
- Skip test_duplicate_clients without bokeh. (:pr:`2553`) Elliott Sales de Andrade
- Add blocked_handlers to servers (:pr:`2556`) Chris White
- Always yield Server.handle_comm coroutine (:pr:`2559`) Tom Augspurger
- Use yaml.safe_load (:pr:`2566`) Matthew Rocklin
- Fetch executables from build root. (:pr:`2551`) Elliott Sales de Andrade
- Fix Torando 6 test failures (:pr:`2570`) Matthew Rocklin
- Fix test_sync_closed_loop (:pr:`2572`) Matthew Rocklin
- Update style to fix recent flake8 update (:pr:`2500`) (:pr:`2509`) Matthew Rocklin
- Fix typo in gen_cluster log message (:pr:`2503`) Loïc Estève
- Allow KeyError when closing event loop (:pr:`2498`) Matthew Rocklin
- Avoid thread testing for TCP ThreadPoolExecutor (:pr:`2510`) Matthew Rocklin
- Find Futures inside SubgraphCallable (:pr:`2505`) Jim Crist
- Avoid AttributeError when closing and sending a message (:pr:`2514`) Matthew Rocklin
- Add deprecation warning to dask_mpi.py (:pr:`2522`) Julia Kent
- Relax statistical profiling test (:pr:`2527`) Matthew Rocklin
- Support alternative --remote-dask-worker SSHCluster() and dask-ssh CLI (:pr:`2526`) Adam Beberg
- Iterate over full list of plugins in transition (:pr:`2518`) Matthew Rocklin
- Create Prometheus Endpoint (:pr:`2499`) Adam Beberg
- Use pytest.importorskip for prometheus test (:pr:`2533`) Matthew Rocklin
- MAINT skip prometheus test when no installed (:pr:`2534`) Olivier Grisel
- Fix intermittent testing failures (:pr:`2535`) Matthew Rocklin
- Avoid using nprocs keyword in dask-ssh if set to one (:pr:`2531`) Matthew Rocklin
- Bump minimum Tornado version to 5.0
- Fix excess threading on missing connections (:pr:`2403`) Daniel Farrell
- Fix typo in doc (:pr:`2457`) Loïc Estève
- Start fewer but larger workers with LocalCluster (:pr:`2452`) Matthew Rocklin
- Check for non-zero
length
first inread
loop (:pr:`2465`) John Kirkham - DOC: Use of local cluster in script (:pr:`2462`) Peter Killick
- DOC/API: Signature for base class write / read (:pr:`2472`) Tom Augspurger
- Support Pytest 4 in Tests (:pr:`2478`) Adam Beberg
- Ensure async behavior in event loop with LocalCluster (:pr:`2484`) Matthew Rocklin
- Fix spurious CancelledError (:pr:`2485`) Loïc Estève
- Properly reset dask.config scheduler and shuffle when closing the client (:pr:`2475`) George Sakkis
- Make it more explicit that resources are per worker. (:pr:`2470`) Loïc Estève
- Remove references to center (:pr:`2488`) Matthew Rocklin
- Expand client clearing timeout to 10s in testing (:pr:`2493`) Matthew Rocklin
- Propagate key keyword in progressbar (:pr:`2492`) Matthew Rocklin
- Use provided cluster's IOLoop if present in Client (:pr:`2494`) Matthew Rocklin
- Clean up LocalCluster logging better in async mode (:pr:`2448`) Matthew Rocklin
- Add short error message if bokeh cannot be imported (:pr:`2444`) Dirk Petersen
- Add optional environment variables to Nanny (:pr:`2431`) Matthew Rocklin
- Make the direct keyword docstring entries uniform (:pr:`2441`) Matthew Rocklin
- Make LocalCluster.close async friendly (:pr:`2437`) Matthew Rocklin
- gather_dep: don't request dependencies we already found out we don't want (:pr:`2428`) tjb900
- Add parameters to Client.run docstring (:pr:`2429`) Matthew Rocklin
- Support coroutines and async-def functions in run/run_scheduler (:pr:`2427`) Matthew Rocklin
- Name threads in ThreadPoolExecutors (:pr:`2408`) Matthew Rocklin
- Serialize numpy.ma.masked objects properly (:pr:`2384`) Jim Crist
- Turn off bokeh property validation in dashboard (:pr:`2387`) Jim Crist
- Fully initialize WorkerState objects (:pr:`2388`) Jim Crist
- Fix typo in scheduler docstring (:pr:`2393`) Russ Bubley
- DOC: fix typo in distributed.worker.Worker docstring (:pr:`2395`) Loïc Estève
- Remove clients and workers from event log after removal (:pr:`2394`) tjb900
- Support msgpack 0.6.0 by providing length keywords (:pr:`2399`) tjb900
- Use async-await on large messages test (:pr:`2404`) Matthew Rocklin
- Fix race condition in normalize_collection (:pr:`2386`) Jim Crist
- Fix redict collection after HighLevelGraph fix upstream (:pr:`2413`) Matthew Rocklin
- Add a blocking argument to Lock.acquire() (:pr:`2412`) Stephan Hoyer
- Fix long traceback test (:pr:`2417`) Matthew Rocklin
- Update x509 certificates to current OpenSSL standards. (:pr:`2418`) Diane Trout
- Fixed the 404 error on the Scheduler Dashboard homepage (:pr:`2361`) Michael Wheeler
- Consolidate two Worker classes into one (:pr:`2363`) Matthew Rocklin
- Avoid warnings in pyarrow and msgpack (:pr:`2364`) Matthew Rocklin
- Avoid race condition in Actor's Future (:pr:`2374`) Matthew Rocklin
- Support missing packages keyword in Client.get_versions (:pr:`2379`) Matthew Rocklin
- Fixup serializing masked arrays (:pr:`2373`) Jim Crist
- Add support for Bokeh 1.0 (:pr:`2348`) (:pr:`2356`) Matthew Rocklin
- Fix regression that dropped support for Tornado 4 (:pr:`2353`) Roy Wedge
- Avoid deprecation warnings (:pr:`2355`) (:pr:`2357`) Matthew Rocklin
- Fix typo in worker documentation (:pr:`2349`) Tom Rochette
- Use tornado's builtin AnyThreadLoopEventPolicy (:pr:`2326`) Matthew Rocklin
- Adjust TLS tests for openssl 1.1 (:pr:`2331`) Marius van Niekerk
- Avoid setting event loop policy if within Jupyter notebook server (:pr:`2343`) Matthew Rocklin
- Add preload script to conf (:pr:`2325`) Guillaume Eynard-Bontemps
- Add serializer for Numpy masked arrays (:pr:`2335`) Peter Killick
- Use psutil.Process.oneshot (:pr:`2339`) NotSqrt
- Use worker SSL context when getting client from worker. (:pr:`2301`) Anonymous
- Remove Joblib Dask Backend from codebase (:pr:`2298`) Matthew Rocklin
- Include worker tls protocol in Scheduler.restart (:pr:`2295`) Matthew Rocklin
- Adapt to new Bokeh selection for 1.0 (:pr:`2292`) Matthew Rocklin
- Add explicit retry method to Future and Client (:pr:`2299`) Matthew Rocklin
- Point to main worker page in bokeh links (:pr:`2300`) Matthew Rocklin
- Limit concurrency when gathering many times (:pr:`2303`) Matthew Rocklin
- Add tls_cluster pytest fixture (:pr:`2302`) Matthew Rocklin
- Convert ConnectionPool.open and active to properties (:pr:`2304`) Matthew Rocklin
- change export_tb to format_tb (:pr:`2306`) Eric Ma
- Redirect joblib page to dask-ml (:pr:`2307`) Matthew Rocklin
- Include unserializable object in error message (:pr:`2310`) Matthew Rocklin
- Import Mapping, Iterator, Set from collections.abc in Python 3 (:pr:`2315`) Gaurav Sheni
- Extend Client.scatter docstring (:pr:`2320`) Eric Ma
- Update for new flake8 (:pr:`2321`) Matthew Rocklin
- Err in dask serialization if not a NotImplementedError (:pr:`2251`) Matthew Rocklin
- Protect against key missing from priority in GraphLayout (:pr:`2259`) Matthew Rocklin
- Do not pull data twice in Client.gather (:pr:`2263`) Adam Klein
- Add pytest fixture for cluster tests (:pr:`2262`) Matthew Rocklin
- Cleanup bokeh callbacks (:pr:`2261`) (:pr:`2278`) Matthew Rocklin
- Fix bokeh error for memory_limit=None (:pr:`2255`) Brett Naul
- Place large keywords into task graph in Client.map (:pr:`2281`) Matthew Rocklin
- Remove redundant blosc threading code from protocol.numpy (:pr:`2284`) Mike Gevaert
- Add ncores to workertable (:pr:`2289`) Matthew Rocklin
- Support upload_file on files with no extension (:pr:`2290`) Matthew Rocklin
- Discard dependent rather than remove (:pr:`2250`) Matthew Rocklin
- Use dask_sphinx_theme Matthew Rocklin
- Drop the Bokeh index page (:pr:`2241`) John Kirkham
- Revert change to keep link relative (:pr:`2242`) Matthew Rocklin
- docs: Fix broken AWS link in setup.rst file (:pr:`2240`) Vladyslav Moisieienkov
- Return cancelled futures in as_completed (:pr:`2233`) Chris White
- Raise informative error when mixing futures between clients (:pr:`2227`) Matthew Rocklin
- add byte_keys to unpack_remotedata call (:pr:`2232`) Matthew Rocklin
- Add documentation for gist/rawgit for get_task_stream (:pr:`2236`) Matthew Rocklin
- Quiet Client.close by waiting for scheduler stop signal (:pr:`2237`) Matthew Rocklin
- Display system graphs nicely on different screen sizes (:pr:`2239`) Derek Ludwig
- Mutate passed in workers dict in TaskStreamPlugin.rectangles (:pr:`2238`) Matthew Rocklin
- Add direct_to_workers to Client Matthew Rocklin
- Add Scheduler.proxy to workers Matthew Rocklin
- Implement Actors Matthew Rocklin
- Fix tooltip (:pr:`2168`) Loïc Estève
- Fix scale / avoid returning coroutines (:pr:`2171`) Joe Hamman
- Clarify dask-worker --nprocs (:pr:`2173`) Yu Feng
- Concatenate all bytes of small messages in TCP comms (:pr:`2172`) Matthew Rocklin
- Add dashboard_link property (:pr:`2176`) Jacob Tomlinson
- Always offload to_frames (:pr:`2170`) Matthew Rocklin
- Warn if desired port is already in use (:pr:`2191`) (:pr:`2199`) Matthew Rocklin
- Add profile page for event loop thread (:pr:`2144`) Matthew Rocklin
- Use dispatch for dask serialization, also add sklearn, pytorch (:pr:`2175`) Matthew Rocklin
- Handle corner cases with busy signal (:pr:`2182`) Matthew Rocklin
- Check self.dependencies when looking at tasks in memory (:pr:`2196`) Matthew Rocklin
- Add ability to log additional custom metrics from each worker (:pr:`2169`) Loïc Estève
- Fix formatting when port is a tuple (:pr:`2204`) Loïc Estève
- Describe what ZeroMQ is (:pr:`2211`) Mike DePalatis
- Tiny typo fix (:pr:`2214`) Anderson Banihirwe
- Add Python 3.7 to travis.yml (:pr:`2203`) Matthew Rocklin
- Add plot= keyword to get_task_stream (:pr:`2198`) Matthew Rocklin
- Add support for optional versions in Client.get_versions (:pr:`2216`) Matthew Rocklin
- Add routes for solo bokeh figures in dashboard (:pr:`2185`) Matthew Rocklin
- Be resilient to missing dep after busy signal (:pr:`2217`) Matthew Rocklin
- Use CSS Grid to layout status page on the dashboard (:pr:`2213`) Derek Ludwig and Luke Canavan
- Fix deserialization of queues on main ioloop thread (:pr:`2221`) Matthew Rocklin
- Add a worker initialization function (:pr:`2201`) Guillaume Eynard-Bontemps
- Collapse navbar in dashboard (:pr:`2223`) Luke Canavan
- Add worker_class= keyword to Nanny to support different worker types (:pr:`2147`) Martin Durant
- Cleanup intermittent worker failures (:pr:`2152`) (:pr:`2146`) Matthew Rocklin
- Fix msgpack PendingDeprecationWarning for encoding='utf-8' (:pr:`2153`) Olivier Grisel
- Make bokeh coloring deterministic using hash function (:pr:`2143`) Matthew Rocklin
- Allow client to query the task stream plot (:pr:`2122`) Matthew Rocklin
- Use PID and counter in thread names (:pr:`2084`) (:pr:`2128`) Dror Birkman
- Test that worker restrictions are cleared after cancellation (:pr:`2107`) Matthew Rocklin
- Expand resources in graph_to_futures (:pr:`2131`) Matthew Rocklin
- Add custom serialization support for pyarrow (:pr:`2115`) Dave Hirschfeld
- Update dask-scheduler cli help text for preload (:pr:`2120`) Matt Nicolls
- Added another nested parallelism test (:pr:`1710`) Tom Augspurger
- insert newline by default after TextProgressBar (:pr:`1976`) Phil Tooley
- Retire workers from scale (:pr:`2104`) Matthew Rocklin
- Allow worker to refuse data requests with busy signal (:pr:`2092`) Matthew Rocklin
- Don't forget released keys (:pr:`2098`) Matthew Rocklin
- Update example for stopping a worker (:pr:`2088`) John Kirkham
- removed hardcoded value of memory terminate fraction from a log message (:pr:`2096`) Bartosz Marcinkowski
- Adjust worker doc after change in config file location and treatment (:pr:`2094`) Aurélien Ponte
- Prefer gathering data from same host (:pr:`2090`) Matthew Rocklin
- Handle exceptions on deserialized comm with text error (:pr:`2093`) Matthew Rocklin
- Fix typo in docstring (:pr:`2087`) Loïc Estève
- Provide communication context to serialization functions (:pr:`2054`) Matthew Rocklin
- Allow name to be explicitly passed in publish_dataset (:pr:`1995`) Marius van Niekerk
- Avoid accessing Worker.scheduler_delay around yield point (:pr:`2074`) Matthew Rocklin
- Support TB and PB in format bytes (:pr:`2072`) Matthew Rocklin
- Add test for as_completed for loops in Python 2 (:pr:`2071`) Matthew Rocklin
- Allow adaptive to exist without a cluster (:pr:`2064`) Matthew Rocklin
- Have worker data transfer wait until recipient acknowledges (:pr:`2052`) Matthew Rocklin
- Support async def functions in Client.sync (:pr:`2070`) Matthew Rocklin
- Add asynchronous parameter to docstring of LocalCluster Matthew Rocklin
- Normalize address before comparison (:pr:`2066`) Tom Augspurger
- Use ConnectionPool for Worker.scheduler Matthew Rocklin
- Avoid reference cycle in str_graph Matthew Rocklin
- Pull data outside of while loop in gather (:pr:`2059`) Matthew Rocklin
- Overhaul configuration (:pr:`1948`) Matthew Rocklin
- Replace get= keyword with scheduler= (:pr:`1959`) Matthew Rocklin
- Use tuples in msgpack (:pr:`2000`) Matthew Rocklin and Marius van Niekerk
- Unify handling of high-volume connections (:pr:`1970`) Matthew Rocklin
- Automatically scatter large arguments in joblib connector (:pr:`2020`) (:pr:`2030`) Olivier Grisel
- Turn click Python 3 locales failure into a warning (:pr:`2001`) Matthew Rocklin
- Rely on dask implementation of sizeof (:pr:`2042`) Matthew Rocklin
- Replace deprecated workers.iloc with workers.values() (:pr:`2013`) Grant Jenks
- Introduce serialization families (:pr:`1912`) Matthew Rocklin
- Add PubSub (:pr:`1999`) Matthew Rocklin
- Add Dask stylesheet to documentation Matthew Rocklin
- Avoid recomputation on partially-complete results (:pr:`1840`) Matthew Rocklin
- Use sys.prefix in popen for testing (:pr:`1954`) Matthew Rocklin
- Include yaml files in manifest Matthew Rocklin
- Use self.sync so Client.processing works in asynchronous context (:pr:`1962`) Henry Doupe
- Fix bug with bad repr on closed client (:pr:`1965`) Matthew Rocklin
- Parse --death-timeout keyword in dask-worker (:pr:`1967`) Matthew Rocklin
- Support serializers in BatchedSend (:pr:`1964`) Matthew Rocklin
- Use normal serialization mechanisms to serialize published datasets (:pr:`1972`) Matthew Rocklin
- Add security support to LocalCluster. (:pr:`1855`) Marius van Niekerk
- add ConnectionPool.remove method (:pr:`1977`) Tony Lorenzo
- Cleanly close workers when scheduler closes (:pr:`1981`) Matthew Rocklin
- Add .pyz support in upload_file (:pr:`1781`) @bmaisson
- add comm to packages (:pr:`1980`) Matthew Rocklin
- Replace dask.set_options with dask.config.set Matthew Rocklin
- Exclude versions of sortedcontainers which do not have .iloc. (:pr:`1993`) Russ Bubley
- Exclude gc statistics under PyPy (:pr:`1997`) Marius van Niekerk
- Manage recent config and dataframe changes in dask (:pr:`2009`) Matthew Rocklin
- Cleanup lingering clients in tests (:pr:`2012`) Matthew Rocklin
- Use timeouts during Client._ensure_connected (:pr:`2011`) Martin Durant
- Avoid reference cycle in joblib backend (:pr:`2014`) Matthew Rocklin, also Olivier Grisel
- DOC: fixed test example (:pr:`2017`) Tom Augspurger
- Add worker_key parameter to Adaptive (:pr:`1992`) Matthew Rocklin
- Prioritize tasks with their true keys, before stringifying (:pr:`2006`) Matthew Rocklin
- Serialize worker exceptions through normal channels (:pr:`2016`) Matthew Rocklin
- Include exception in progress bar (:pr:`2028`) Matthew Rocklin
- Avoid logging orphaned futures in All (:pr:`2008`) Matthew Rocklin
- Don't use spill-to-disk dictionary if we're not spilling to disk Matthew Rocklin
- Only avoid recomputation if key exists (:pr:`2036`) Matthew Rocklin
- Use client connection and serialization arguments in progress (:pr:`2035`) Matthew Rocklin
- Rejoin worker client on closing context manager (:pr:`2041`) Matthew Rocklin
- Avoid forgetting erred tasks when losing dependencies (:pr:`2047`) Matthew Rocklin
- Avoid collisions in graph_layout (:pr:`2050`) Matthew Rocklin
- Avoid recursively calling bokeh callback in profile plot (:pr:`2048`) Matthew Rocklin
- Remove errant print statement (:pr:`1957`) Matthew Rocklin
- Only add reevaluate_occupancy callback once (:pr:`1953`) Tony Lorenzo
- Newline needed for doctest rendering (:pr:`1917`) Loïc Estève
- Support Client._repr_html_ when in async mode (:pr:`1909`) Matthew Rocklin
- Add parameters to dask-ssh command (:pr:`1910`) Irene Rodriguez
- Sanitize get_dataset trace (:pr:`1888`) John Kirkham
- Fix bug where queues would not clean up cleanly (:pr:`1922`) Matthew Rocklin
- Delete cached file safely in upload file (:pr:`1921`) Matthew Rocklin
- Accept KeyError when closing tornado IOLoop in tests (:pr:`1937`) Matthew Rocklin
- Quiet the client and scheduler when gather(..., errors='skip') (:pr:`1936`) Matthew Rocklin
- Clarify couldn't gather keys warning (:pr:`1942`) Kenneth Koski
- Support submit keywords in joblib (:pr:`1947`) Matthew Rocklin
- Avoid use of external resources in bokeh server (:pr:`1934`) Matthew Rocklin
- Drop __contains__ from Datasets (:pr:`1889`) John Kirkham
- Fix bug with queue timeouts (:pr:`1950`) Matthew Rocklin
- Replace msgpack-python by msgpack (:pr:`1927`) Loïc Estève
- Fix numeric environment variable configuration (:pr:`1885`) Joseph Atkins-Kurkish
- support bytearrays in older lz4 library (:pr:`1886`) Matthew Rocklin
- Remove started timeout in nanny (:pr:`1852`) Matthew Rocklin
- Don't log errors in sync (:pr:`1894`) Matthew Rocklin
- downgrade stale lock warning to info logging level (:pr:`1890`) Matthew Rocklin
- Fix
UnboundLocalError
forkey
(:pr:`1900`) John Kirkham - Resolve deployment issues in Python 2 (:pr:`1905`) Matthew Rocklin
- Support retries and priority in Client.get method (:pr:`1902`) Matthew Rocklin
- Add additional attributes to task page if applicable (:pr:`1901`) Matthew Rocklin
- Add count method to as_completed (:pr:`1897`) Matthew Rocklin
- Extend default timeout to 10s (:pr:`1904`) Matthew Rocklin
- Increase default allowable tick time to 3s (:pr:`1854`) Matthew Rocklin
- Handle errant workers when another worker has data (:pr:`1853`) Matthew Rocklin
- Close multiprocessing queue in Nanny to reduce open file descriptors (:pr:`1862`) Matthew Rocklin
- Extend nanny started timeout to 30s, make configurable (:pr:`1865`) Matthew Rocklin
- Comment out the default config file (:pr:`1871`) Matthew Rocklin
- Update to fix bokeh 0.12.15 update errors (:pr:`1872`) Matthew Rocklin
- Downgrade Event Loop unresponsive warning to INFO level (:pr:`1870`) Matthew Rocklin
- Add fifo timeout to control priority generation (:pr:`1828`) Matthew Rocklin
- Add retire_workers API to Client (:pr:`1876`) Matthew Rocklin
- Catch NoSuchProcess error in Nanny.memory_monitor (:pr:`1877`) Matthew Rocklin
- Add uid to nanny queue communications (:pr:`1880`) Matthew Rocklin
- Avoid passing bytearrays to snappy decompression (:pr:`1831`) Matthew Rocklin
- Specify IOLoop in Adaptive (:pr:`1841`) Matthew Rocklin
- Use connect-timeout config value throughout client (:pr:`1839`) Matthew Rocklin
- Support direct= keyword argument in Client.get (:pr:`1845`) Matthew Rocklin
- Add cluster superclass and improve adaptivity (:pr:`1813`) Matthew Rocklin
- Fixup tests and support Python 2 for Tornado 5.0 (:pr:`1818`) Matthew Rocklin
- Fix bug in recreate_error when dependencies are dropped (:pr:`1815`) Matthew Rocklin
- Add worker time to live in Scheduler (:pr:`1811`) Matthew Rocklin
- Scale adaptive based on total_occupancy (:pr:`1807`) Matthew Rocklin
- Support calling compute within worker_client (:pr:`1814`) Matthew Rocklin
- Add percentage to profile plot (:pr:`1817`) Brett Naul
- Overwrite option for remote python in dask-ssh (:pr:`1812`) Sven Kreiss
- Fix bug where we didn't check idle/saturated when stealing (:pr:`1801`) Matthew Rocklin
- Fix bug where client was noisy when scheduler closed unexpectedly (:pr:`1806`) Matthew Rocklin
- Use string-based timedeltas (like
'500 ms'
) everywhere (:pr:`1804`) Matthew Rocklin - Keep logs in scheduler and worker even if silenced (:pr:`1803`) Matthew Rocklin
- Support minimum, maximum, wait_count keywords in Adaptive (:pr:`1797`) Jacob Tomlinson and Matthew Rocklin
- Support async protocols for LocalCluster, replace start= with asynchronous= (:pr:`1798`) Matthew Rocklin
- Avoid restarting workers when nanny waits on scheduler (:pr:`1793`) Matthew Rocklin
- Use
IOStream.read_into()
when available (:pr:`1477`) Antoine Pitrou - Reduce LocalCluster logging threshold from CRITICAL to WARN (:pr:`1785`) Andy Jones
- Add futures_of to API docs (:pr:`1783`) John Kirkham
- Make diagnostics link in client configurable (:pr:`1810`) Matthew Rocklin
- Fixed an uncaught exception in
distributed.joblib
with aLocalCluster
using only threads (:issue:`1775`) Tom Augspurger - Format bytes in info worker page (:pr:`1752`) Matthew Rocklin
- Add pass-through arguments for scheduler/worker --preload modules. (:pr:`1634`) Alex Ford
- Use new LZ4 API (:pr:`1757`) Thrasibule
- Replace dask.optimize with dask.optimization (:pr:`1754`) Matthew Rocklin
- Add graph layout engine and bokeh plot (:pr:`1756`) Matthew Rocklin
- Only expand name with --nprocs if name exists (:pr:`1776`) Matthew Rocklin
- specify IOLoop for stealing PeriodicCallback (:pr:`1777`) Matthew Rocklin
- Fixed distributed.joblib with no processes Tom Augspurger
- Use set.discard to avoid KeyErrors in stealing (:pr:`1766`) Matthew Rocklin
- Avoid KeyError when task has been released during steal (:pr:`1765`) Matthew Rocklin
- Add versions routes to avoid the use of run in Client.get_versions (:pr:`1773`) Matthew Rocklin
- Add write_scheduler_file to Client (:pr:`1778`) Joe Hamman
- Default host to tls:// if tls information provided (:pr:`1780`) Matthew Rocklin
- Refactor scheduler to use TaskState objects rather than dictionaries (:pr:`1594`) Antoine Pitrou
- Plot CPU fraction of total in workers page (:pr:`1624`) Matthew Rocklin
- Use thread CPU time in Throttled GC (:pr:`1625`) Antoine Pitrou
- Fix bug with
memory_limit=None
(:pr:`1639`) Matthew Rocklin - Add futures_of to top level api (:pr:`1646`) Matthew Rocklin
- Warn on serializing large data in Client (:pr:`1636`) Matthew Rocklin
- Fix intermittent windows failure when removing lock file (:pr:`1652`) Antoine Pitrou
- Add diagnosis and logging of poor GC Behavior (:pr:`1635`) Antoine Pitrou
- Add client-scheduler heartbeats (:pr:`1657`) Matthew Rocklin
- Return dictionary of worker info in
retire_workers
(:pr:`1659`) Matthew Rocklin - Ensure dumps_function works with unhashable functions (:pr:`1662`) Matthew Rocklin
- Collect client name ids rom client-name config variable (:pr:`1664`) Matthew Rocklin
- Allow simultaneous use of --name and --nprocs in dask-worker (:pr:`1665`) Matthew Rocklin
- Add support for grouped adaptive scaling and adaptive behavior overrides (:pr:`1632`) Alex Ford
- Share scheduler RPC between worker and client (:pr:`1673`) Matthew Rocklin
- Allow
retries=
in ClientExecutor (:pr:`1672`) @rqx - Improve documentation for get_client and dask.compute examples (:pr:`1638`) Scott Sievert
- Support DASK_SCHEDULER_ADDRESS environment variable in worker (:pr:`1680`) Matthew Rocklin
- Support tuple-keys in retries (:pr:`1681`) Matthew Rocklin
- Use relative links in bokeh dashboard (:pr:`1682`) Matthew Rocklin
- Make message log length configurable, default to zero (:pr:`1691`) Matthew Rocklin
- Deprecate
Client.shutdown
(:pr:`1699`) Matthew Rocklin - Add warning in configuration docs to install pyyaml (:pr:`1701`) Cornelius Riemenschneider
- Handle nested parallelism in distributed.joblib (:pr:`1705`) Tom Augspurger
- Don't wait for Worker.executor to shutdown cleanly when restarting process (:pr:`1708`) Matthew Rocklin
- Add support for user defined priorities (:pr:`1651`) Matthew Rocklin
- Catch and log OSErrors around worker lock files (:pr:`1714`) Matthew Rocklin
- Remove worker prioritization. Coincides with changes to dask.order (:pr:`1730`) Matthew Rocklin
- Use process-measured memory rather than nbytes in Bokeh dashboard (:pr:`1737`) Matthew Rocklin
- Enable serialization of Locks (:pr:`1738`) Matthew Rocklin
- Support Tornado 5 beta (:pr:`1735`) Matthew Rocklin
- Cleanup remote_magic client cache after tests (:pr:`1743`) Min RK
- Allow service ports to be specified as (host, port) (:pr:`1744`) Bruce Merry
- Clear deque handlers after each test (:pr:`1586`) Antoine Pitrou
- Handle deserialization in FutureState.set_error (:pr:`1592`) Matthew Rocklin
- Add process leak checker to tests (:pr:`1596`) Antoine Pitrou
- Customize process title for subprocess (:pr:`1590`) Antoine Pitrou
- Make linting a separate CI job (:pr:`1599`) Antoine Pitrou
- Fix error from get_client() with no global client (:pr:`1595`) Daniel Li
- Remove Worker.host_health, correct WorkerTable metrics (:pr:`1600`) Matthew Rocklin
- Don't mark tasks as suspicious when retire_workers called. Addresses (:pr:`1607`) Russ Bubley
- Do not include processing workers in workers_to_close (:pr:`1609`) Russ Bubley
- Disallow simultaneous scale up and down in Adaptive (:pr:`1608`) Russ Bubley
- Parse bytestrings in --memory-limit (:pr:`1615`) Matthew Rocklin
- Use environment variable for scheduler address if present (:pr:`1610`) Matthew Rocklin
- Fix deprecation warning from logger.warn (:pr:`1616`) Brett Naul
- Wrap
import ssl
statements with try-except block for ssl-crippled environments, (:pr:`1570`) Xander Johnson - Support zero memory-limit in Nanny (:pr:`1571`) Matthew Rocklin
- Avoid PeriodicCallback double starts (:pr:`1573`) Matthew Rocklin
- Add disposable workspace facility (:pr:`1543`) Antoine Pitrou
- Use format_time in task_stream plots (:pr:`1575`) Matthew Rocklin
- Avoid delayed finalize calls in compute (:pr:`1577`) Matthew Rocklin
- Doc fix about secede (:pr:`1583`) Scott Sievert
- Add tracemalloc option when tracking test leaks (:pr:`1585`) Antoine Pitrou
- Add JSON routes to Bokeh server (:pr:`1584`) Matthew Rocklin
- Handle exceptions cleanly in Variables and Queues (:pr:`1580`) Matthew Rocklin
- Drop use of pandas.msgpack (:pr:`1473`) Matthew Rocklin
- Add methods to get/set scheduler metadata Matthew Rocklin
- Add distributed lock Matthew Rocklin
- Add reschedule exception for worker tasks Matthew Rocklin
- Fix
nbytes()
forbytearrays
Matthew Rocklin - Capture scheduler and worker logs Matthew Rocklin
- Garbage collect after data eviction on high worker memory usage (:pr:`1488`) Olivier Grisel
- Add scheduler HTML routes to bokeh server (:pr:`1478`) (:pr:`1514`) Matthew Rocklin
- Add pytest plugin to test for resource leaks (:pr:`1499`) Antoine Pitrou
- Improve documentation for scheduler states (:pr:`1498`) Antoine Pitrou
- Correct warn_if_longer timeout in ThrottledGC (:pr:`1496`) Fabian Keller
- Catch race condition in as_completed on cancelled futures (:pr:`1507`) Matthew Rocklin
- Transactional work stealing (:pr:`1489`) (:pr:`1528`) Matthew Rocklin
- Avoid forkserver in PyPy (:pr:`1509`) Matthew Rocklin
- Add dict access to get/set datasets (:pr:`1508`) Mike DePalatis
- Support Tornado 5 (:pr:`1509`) (:pr:`1512`) (:pr:`1518`) (:pr:`1534`) Antoine Pitrou
- Move thread_state in Dask (:pr:`1523`) Jim Crist
- Use new Dask collections interface (:pr:`1513`) Matthew Rocklin
- Add nanny flag to dask-mpi Matthew Rocklin
- Remove JSON-based HTTP servers Matthew Rocklin
- Avoid doing I/O in repr/str (:pr:`1536`) Matthew Rocklin
- Fix URL for MPI4Py project (:pr:`1546`) Ian Hopkinson
- Allow automatic retries of a failed task (:pr:`1524`) Antoine Pitrou
- Clean and accelerate tests (:pr:`1548`) (:pr:`1549`) (:pr:`1552`) (:pr:`1553`) (:pr:`1560`) (:pr:`1564`) Antoine Pitrou
- Move HDFS functionality to the hdfs3 library (:pr:`1561`) Jim Crist
- Fix bug when using events page with no events (:pr:`1562`) @rbubley
- Improve diagnostic naming of tasks within tuples (:pr:`1566`) Kelvyn Yang
- Handle None case in profile.identity (:pr:`1456`)
- Asyncio rewrite (:pr:`1458`)
- Add rejoin function partner to secede (:pr:`1462`)
- Nested compute (:pr:`1465`)
- Use LooseVersion when comparing Bokeh versions (:pr:`1470`)
- as_completed doesn't block on cancelled futures (:pr:`1436`)
- Notify waiting threads/coroutines on cancellation (:pr:`1438`)
- Set Future(inform=True) as default (:pr:`1437`)
- Rename Scheduler.transition_story to story (:pr:`1445`)
- Future uses default client by default (:pr:`1449`)
- Add keys= keyword to Client.call_stack (:pr:`1446`)
- Add get_current_task to worker (:pr:`1444`)
- Ensure that Client remains asynchronous before ioloop starts (:pr:`1452`)
- Remove "click for worker page" in bokeh plot (:pr:`1453`)
- Add Client.current() (:pr:`1450`)
- Clean handling of restart timeouts (:pr:`1442`)
- Fix tool issues with TaskStream plot (:pr:`1425`)
- Move profile module to top level (:pr:`1423`)
- Avoid storing messages in message log (:pr:`1361`)
- fileConfig does not disable existing loggers (:pr:`1380`)
- Offload upload_file disk I/O to separate thread (:pr:`1383`)
- Add missing SSLContext (:pr:`1385`)
- Collect worker thread information from sys._curent_frames (:pr:`1387`)
- Add nanny timeout (:pr:`1395`)
- Restart worker if memory use goes above 95% (:pr:`1397`)
- Track workers memory use with psutil (:pr:`1398`)
- Track scheduler delay times in workers (:pr:`1400`)
- Add time slider to profile plot (:pr:`1403`)
- Change memory-limit keyword to refer to maximum number of bytes (:pr:`1405`)
- Add
cancel(force=)
keyword (:pr:`1408`)
- Silently pass on cancelled futures in as_completed (:pr:`1366`)
- Fix unicode keys error in Python 2 (:pr:`1370`)
- Support numeric worker names
- Add dask-mpi executable (:pr:`1367`)
- Clean up forgotten keys in fire-and-forget workloads (:pr:`1250`)
- Handle missing extensions (:pr:`1263`)
- Allow recreate_exception on persisted collections (:pr:`1253`)
- Add asynchronous= keyword to blocking client methods (:pr:`1272`)
- Restrict to horizontal panning in bokeh plots (:pr:`1274`)
- Rename client.shutdown to client.close (:pr:`1275`)
- Avoid blocking on event loop (:pr:`1270`)
- Avoid cloudpickle errors for Client.get_versions (:pr:`1279`)
- Yield on Tornado IOStream.write futures (:pr:`1289`)
- Assume async behavior if inside a sync statement (:pr:`1284`)
- Avoid error messages on closing (:pr:`1297`), (:pr:`1296`) (:pr:`1318`) (:pr:`1319`)
- Add timeout= keyword to get_client (:pr:`1290`)
- Respect timeouts when restarting (:pr:`1304`)
- Clean file descriptor and memory leaks in tests (:pr:`1317`)
- Deprecate Executor (:pr:`1302`)
- Add timeout to ThreadPoolExecutor.shutdown (:pr:`1330`)
- Clean up AsyncProcess handling (:pr:`1324`)
- Allow unicode keys in Python 2 scheduler (:pr:`1328`)
- Avoid leaking stolen data (:pr:`1326`)
- Improve error handling on failed nanny starts (:pr:`1337`), (:pr:`1331`)
- Make Adaptive more flexible
- Support
--contact-address
and--listen-address
in worker (:pr:`1278`) - Remove old dworker, dscheduler executables (:pr:`1355`)
- Exit workers if nanny process fails (:pr:`1345`)
- Auto pep8 and flake (:pr:`1353`)
- Multi-threading safety (:pr:`1191`), (:pr:`1228`), (:pr:`1229`)
- Improve handling of byte counting (:pr:`1198`) (:pr:`1224`)
- Add get_client, secede functions, refactor worker-client relationship (:pr:`1201`)
- Allow logging configuration using logging.dictConfig() (:pr:`1206`) (:pr:`1211`)
- Offload serialization and deserialization to separate thread (:pr:`1218`)
- Support fire-and-forget tasks (:pr:`1221`)
- Support bytestrings as keys (for Julia) (:pr:`1234`)
- Resolve testing corner-cases (:pr:`1236`), (:pr:`1237`), (:pr:`1240`), (:pr:`1241`), (:pr:`1242`), (:pr:`1244`)
- Automatic use of scatter/gather(direct=True) in more cases (:pr:`1239`)
- Remove Python 3.4 testing from travis-ci (:pr:`1157`)
- Remove ZMQ Support (:pr:`1160`)
- Fix memoryview nbytes issue in Python 2.7 (:pr:`1165`)
- Re-enable counters (:pr:`1168`)
- Improve scheduler.restart (:pr:`1175`)
- Reevaluate worker occupancy periodically during scheduler downtime (:pr:`1038`) (:pr:`1101`)
- Add
AioClient
asyncio-compatible client API (:pr:`1029`) (:pr:`1092`) (:pr:`1099`) - Update Keras serializer (:pr:`1067`)
- Support TLS/SSL connections for security (:pr:`866`) (:pr:`1034`)
- Always create new worker directory when passed
--local-directory
(:pr:`1079`) - Support pre-scattering data when using joblib frontend (:pr:`1022`)
- Make workers more robust to failure of
sizeof
function (:pr:`1108`) and writing to disk (:pr:`1096`) - Add
is_empty
andupdate
methods toas_completed
(:pr:`1113`) - Remove
_get
coroutine and replace withget(..., sync=False)
(:pr:`1109`) - Improve API compatibility with async/await syntax (:pr:`1115`) (:pr:`1124`)
- Add distributed Queues (:pr:`1117`) and shared Variables (:pr:`1128`) to enable inter-client coordination
- Support direct client-to-worker scattering and gathering (:pr:`1130`) as well as performance enhancements when scattering data
- Style improvements for bokeh web dashboards (:pr:`1126`) (:pr:`1141`) as well as a removal of the external bokeh process
- HTML reprs for Future and Client objects (:pr:`1136`)
- Support nested collections in client.compute (:pr:`1144`)
- Use normal client API in asynchronous mode (:pr:`1152`)
- Remove old distributed.collections submodule (:pr:`1153`)
- Add bokeh template files to MANIFEST (:pr:`1063`)
- Don't set worker_client.get as default get (:pr:`1061`)
- Clean up logging on Client().shutdown() (:pr:`1055`)
- Support
async with Client
syntax (:pr:`1053`) - Use internal bokeh server for default diagnostics server (:pr:`1047`)
- Improve styling of bokeh plots when empty (:pr:`1046`) (:pr:`1037`)
- Support efficient serialization for sparse arrays (:pr:`1040`)
- Prioritize newly arrived work in worker (:pr:`1035`)
- Prescatter data with joblib backend (:pr:`1022`)
- Make client.restart more robust to worker failure (:pr:`1018`)
- Support preloading a module or script in dask-worker or dask-scheduler processes (:pr:`1016`)
- Specify network interface in command line interface (:pr:`1007`)
- Client.scatter supports a single element (:pr:`1003`)
- Use blosc compression on all memoryviews passing through comms (:pr:`998`)
- Add concurrent.futures-compatible Executor (:pr:`997`)
- Add as_completed.batches method and return results (:pr:`994`) (:pr:`971`)
- Allow worker_clients to optionally stay within the thread pool (:pr:`993`)
- Add bytes-stored and tasks-processing diagnostic histograms (:pr:`990`)
- Run supports non-msgpack-serializable results (:pr:`965`)
- Use inproc transport in LocalCluster (:pr:`919`)
- Add structured and queryable cluster event logs (:pr:`922`)
- Use connection pool for inter-worker communication (:pr:`935`)
- Robustly shut down spawned worker processes at shutdown (:pr:`928`)
- Worker death timeout (:pr:`940`)
- More visual reporting of exceptions in progressbar (:pr:`941`)
- Render disk and serialization events to task stream visual (:pr:`943`)
- Support async for / await protocol (:pr:`952`)
- Ensure random generators are re-seeded in worker processes (:pr:`953`)
- Upload sourcecode as zip module (:pr:`886`)
- Replay remote exceptions in local process (:pr:`894`)
- First come first served priorities on client submissions (:pr:`840`)
- Can specify Bokeh internal ports (:pr:`850`)
- Allow stolen tasks to return from either worker (:pr:`853`), (:pr:`875`)
- Add worker resource constraints during execution (:pr:`857`)
- Send small data through Channels (:pr:`858`)
- Better estimates for SciPy sparse matrix memory costs (:pr:`863`)
- Avoid stealing long running tasks (:pr:`873`)
- Maintain fortran ordering of NumPy arrays (:pr:`876`)
- Add
--scheduler-file
keyword to dask-scheduler (:pr:`877`) - Add serializer for Keras models (:pr:`878`)
- Support uploading modules from zip files (:pr:`886`)
- Improve titles of Bokeh dashboards (:pr:`895`)
- Fix a bug where arrays with large dtypes or shapes were being improperly compressed (:pr:`830` :pr:`832` :pr:`833`)
- Extend
as_completed
to accept new futures during iteration (:pr:`829`) - Add
--nohost
keyword todask-ssh
startup utility (:pr:`827`) - Support scheduler shutdown of remote workers, useful for adaptive clusters (:pr: 811 :pr:`816` :pr:`821`)
- Add
Client.run_on_scheduler
method for running debug functions on the scheduler (:pr:`808`)
- Make compatible with Bokeh 0.12.4 (:pr:`803`)
- Avoid compressing arrays if not helpful (:pr:`777`)
- Optimize inter-worker data transfer (:pr:`770`) (:pr:`790`)
- Add --local-directory keyword to worker (:pr:`788`)
- Enable workers to arrive to the cluster with their own data. Useful if a worker leaves and comes back (:pr:`785`)
- Resolve thread safety bug when using local_client (:pr:`802`)
- Resolve scheduling issues in worker (:pr:`804`)
- Major Worker refactor (:pr:`704`)
- Major Scheduler refactor (:pr:`717`) (:pr:`722`) (:pr:`724`) (:pr:`742`) (:pr:`743`
- Add
check
(default isFalse
) option toClient.get_versions
to raise if the versions don't match on client, scheduler & workers (:pr:`664`) Future.add_done_callback
executes in separate thread (:pr:`656`)- Clean up numpy serialization (:pr:`670`)
- Support serialization of Tornado v4.5 coroutines (:pr:`673`)
- Use CPickle instead of Pickle in Python 2 (:pr:`684`)
- Use Forkserver rather than Fork on Unix in Python 3 (:pr:`687`)
- Support abstract resources for per-task constraints (:pr:`694`) (:pr:`720`) (:pr:`737`)
- Add TCP timeouts (:pr:`697`)
- Add embedded Bokeh server to workers (:pr:`709`) (:pr:`713`) (:pr:`738`)
- Add embedded Bokeh server to scheduler (:pr:`724`) (:pr:`736`) (:pr:`738`)
- Add more precise timers for Windows (:pr:`713`)
- Add Versioneer (:pr:`715`)
- Support inter-client channels (:pr:`729`) (:pr:`749`)
- Scheduler Performance improvements (:pr:`740`) (:pr:`760`)
- Improve load balancing and work stealing (:pr:`747`) (:pr:`754`) (:pr:`757`)
- Run Tornado coroutines on workers
- Avoid slow sizeof call on Pandas dataframes (:pr:`758`)
- Remove custom Bokeh export tool that implicitly relied on nodejs (:pr:`655`)
- Clean up scheduler logging (:pr:`657`)
- Support more numpy dtypes in custom serialization, (:pr:`627`), (:pr:`630`), (:pr:`636`)
- Update Bokeh plots (:pr:`628`)
- Improve spill to disk heuristics (:pr:`633`)
- Add Export tool to Task Stream plot
- Reverse frame order in loads for very many frames (:pr:`651`)
- Add timeout when waiting on write (:pr:`653`)
- Add
Client.get_versions()
function to return software and package information from the scheduler, workers, and client (:pr:`595`) - Improved windows support (:pr:`577`) (:pr:`590`) (:pr:`583`) (:pr:`597`)
- Clean up rpc objects explicitly (:pr:`584`)
- Normalize collections against known futures (:pr:`587`)
- Add key= keyword to map to specify keynames (:pr:`589`)
- Custom data serialization (:pr:`606`)
- Refactor the web interface (:pr:`608`) (:pr:`615`) (:pr:`621`)
- Allow user-supplied Executor in Worker (:pr:`609`)
- Pass Worker kwargs through LocalCluster
- Schedulers can retire workers cleanly
- Add
Future.add_done_callback
forconcurrent.futures
compatibility - Update web interface to be consistent with Bokeh 0.12.3
- Close streams explicitly, avoiding race conditions and supporting more robust restarts on Windows.
- Improved shuffled performance for dask.dataframe
- Add adaptive allocation cluster manager
- Reduce administrative overhead when dealing with many workers
dask-ssh --log-directory .
no longer errors- Microperformance tuning for the scheduler
- Revert dask_worker to use fork rather than subprocess by default
- Scatter retains type information
- Bokeh always uses subprocess rather than spawn
- Fix critical Windows error with dask_worker executable
- Rename Executor to Client (:pr:`492`)
- Add
--memory-limit
option todask-worker
, enabling spill-to-disk behavior when running out of memory (:pr:`485`) - Add
--pid-file
option to dask-worker and--dask-scheduler
(:pr:`496`) - Add
upload_environment
function to distribute conda environments. This is experimental, undocumented, and may change without notice. (:pr:`494`) - Add
workers=
keyword argument toClient.compute
andClient.persist
, supporting location-restricted workloads with Dask collections (:pr:`484`) - Add
upload_environment
function to distribute conda environments. This is experimental, undocumented, and may change without notice. (:pr:`494`)- Add optional
dask_worker=
keyword toclient.run
functions that gets provided the worker or nanny object - Add
nanny=False
keyword toClient.run
, allowing for the execution of arbitrary functions on the nannies as well as normal workers
- Add optional
This release adds some new features and removes dead code
- Publish and share datasets on the scheduler between many clients (:pr:`453`). See :doc:`publish`.
- Launch tasks from other tasks (experimental) (:pr:`471`). See :doc:`task-launch`.
- Remove unused code, notably the
Center
object and older client functions (:pr:`478`) Executor()
andLocalCluster()
is now robust to Bokeh's absence (:pr:`481`)- Removed s3fs and boto3 from requirements. These have moved to Dask.
This release is largely a bugfix release, recovering from the previous large refactor.
- Fixes from previous refactor
- Ensure idempotence across clients
- Stress test losing scattered data permanently
- IPython fixes
- Add
start_ipython_scheduler
method to Executor - Add
%remote
magic for workers - Clean up code and tests
- Add
- Pool connects to maintain reuse and reduce number of open file handles
- Re-implement work stealing algorithm
- Support cancellation of tuple keys, such as occur in dask.arrays
- Start synchronizing against worker data that may be superfluous
- Improve bokeh plots styling
- Add memory plot tracking number of bytes
- Make the progress bars more compact and align colors
- Add workers/ page with workers table, stacks/processing plot, and memory
- Add this release notes document
This release was largely a refactoring release. Internals were changed significantly without many new features.
- Major refactor of the scheduler to use transitions system
- Tweak protocol to traverse down complex messages in search of large bytestrings
- Add dask-submit and dask-remote
- Refactor HDFS writing to align with changes in the dask library
- Executor reconnects to scheduler on broken connection or failed scheduler
- Support sklearn.external.joblib as well as normal joblib