Skip to content

WIP: Add option to enable RMM logging #3924

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
jakirkham wants to merge 3 commits into
base: main
Choose a base branch
from
Draft

WIP: Add option to enable RMM logging #3924

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
76be1bc
Add option to enable RMM logging
jakirkham Jun 23, 2020
c5f3330
Fix `logging` type in config schema
jakirkham Jun 24, 2020
f500557
fix up schema
quasiben Jul 1, 2020
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 6 additions & 1 deletion distributed/comm/ucx.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -101,10 +101,15 @@ def device_array(n):
)

pool_size_str = dask.config.get("rmm.pool-size")
enable_logging = dask.config.get("rmm.logging")
if pool_size_str is not None:
pool_size = parse_bytes(pool_size_str)
rmm.reinitialize(
pool_allocator=True, managed_memory=False, initial_pool_size=pool_size
pool_allocator=True,
managed_memory=False,
initial_pool_size=pool_size,
logging=enable_logging,
log_file_name="", # Something more useful here ;)

@jakirkham jakirkham Jul 7, 2020

Copy link
Copy Markdown
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We still need to fill this in with something. This is a bit tricky to get right as we would presumably want a different location depending on whether we are setting this on a worker, on the scheduler, or potentially on the client. However I'm not sure how we access that information from a context that is not inside one of those classes. Welcome thoughts on that though 🙂

@pentschev pentschev Jul 7, 2020

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Maybe we should force the user to pass a path+filename prefix if logging is enabled and append the PID to it?

)


Expand Down
4 changes: 4 additions & 0 deletions distributed/distributed-schema.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -770,6 +770,10 @@ properties:
- "null"
description:
The size of the memory pool in bytes
logging:
type: boolean
description:
Whether to enable logging with RMM
ucx:
type: object
description: |
Expand Down
1 change: 1 addition & 0 deletions distributed/distributed.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -172,6 +172,7 @@ distributed:
pdb-on-err: False # enter debug mode on scheduling error
rmm:
pool-size: null
logging: False
ucx:
tcp: null # enable tcp
nvlink: null # enable cuda_ipc
Expand Down