Skip to content

v0.6.0

Pre-release
Pre-release

Choose a tag to compare

@github-actions github-actions released this 10 Nov 11:31
· 25 commits to dev since this release
Immutable release. Only release title and notes can be modified.
v0.6.0
55c4d15

proTES now runs on FOCA 0.22 and foca-plugins 0.3.0, and discovers TES
instances in Cloud Registry 1.0.0rc2 again, also those that serve the TES API
at another path. Admins can limit the tasks of each user and delete finished
tasks after a number of days, both opt-in. Responses of TES instances and pages
of the view FULL are bounded in size, tasks stuck in CANCELING end after a
time, and service infos of TES instances are read without credentials. Several
changes are breaking: read the upgrade steps first. See the documentation for
details.

Breaking changes

  • proTES runs on FOCA 0.22.0 and its image: ${...} in config files references
    an environment variable, responses have the header X-Request-ID (generated,
    clients' values ignored unless server.request_id.trust_incoming), query
    parameters on operations without parameters get 400, the Celery worker
    stops on invalid access control settings, MongoDB clients time out after 5
    seconds, and refusals of the plugin gateway and log entries of requests have
    other messages; plugins that cap foca<0.22 no longer resolve against the
    image (see the upgrade
    steps
    )
    (#433, #434; #462, #463)
  • The image contains the release train foca-plugins 0.3.0 (plugins 0.1.3)
    instead of 0.2.0 (#467, #471)
  • Service infos of TES instances are read without credentials; TES instances
    that refuse anonymous reads count as TES 1.0 unless they have the label
    tes.version or custom.tes.probe_with_token is set (#436, #448)
  • Tasks still CANCELING custom.controllers.post_task.polling.max_cancel_time
    seconds (default 3600) after their cancel are set to SYSTEM_ERROR and no
    longer tracked (#437, #450)
  • Responses of TES instances are read up to custom.tes.max_response_size
    bytes (default 10 MiB, was 64 MiB for some): larger task logs are left out,
    and a larger response to a task creation leaves the task in SYSTEM_ERROR
    (#439, #452)
  • Pages of the view FULL also end before their tasks exceed
    custom.controllers.list_tasks.max_page_bytes_full bytes (10 MiB, 4 MiB in
    the Helm chart), so they may hold fewer tasks (#455, #458)
  • Docker Compose runs the images of the Helm chart, pinned to RabbitMQ 4.3.7
    and MongoDB 7.0.45; RabbitMQ may need a new volume (#435, #440)

Features

  • Opt-in limits of the unfinished and stored tasks per user
    (custom.controllers.post_task.limits) (#285, #456)
  • Opt-in retention: the Celery workers delete finished tasks
    custom.retention.finished_days days after they finished, leaving tombstones
    (#286, #457)
  • Discovered TES instances whose registry URL serves the TES API (e.g.,
    TESP-API at /v1) are used at that URL, without an entry of their own with
    api_path (#438, #451)
  • Discovery works with Cloud Registry 1.0.0rc2, whose GET /verifications
    names services by serviceId (#462, #463)
  • Responses to requests that fail the validation of the API specification
    keep the validator's reason as detail (exceptions.validation_reasons),
    and gain type, code and the invalid fields (errors) where these are
    known (#462, #463)

Security

  • By default, proTES' access token and Basic credentials are no longer sent
    with reads of service infos, so TES instances that tasks don't go to don't
    get them; custom.tes.probe_with_token: true sends them as before
    (#436, #448)
  • Requests to TES instances no longer take credentials from a .netrc file of
    the environment (#449, #454)
  • The size limit of responses of TES instances bounds the memory that reading
    them takes, also for compressed responses (#439, #452)

Fixes

  • Tasks whose TES instance never confirms a cancel are no longer tracked
    forever (#437, #450)
  • One page of the view FULL with large tasks could exceed the memory limit
    of the app (#455, #458)

Documentation

  • The upgrade path from 0.5.0 (#468, #472)

Maintenance

  • Moved to FOCA 0.21 (#433, #434) and 0.22 (#462, #463)
  • CI runs tasks on TESK in a k3s cluster, in the job Integration (TESK)
    (#423, #432)
  • Tests read the config with get_config() and cover
    exceptions.validation_reasons: false (#441, #443; #469, #470)
  • The release workflow verifies the release tag before running its scripts
    (#446, #447)
  • Removed unused models and ignore entries (#444, #445); Dependabot PRs get
    the label type: dependencies (#460, #461)

Commits

  • test: run against TESK in a k3s cluster by @uniqueg in #432
  • build: move to FOCA 0.21 by @uniqueg in #434
  • build(deps): bump the docker group across 2 directories with 1 update by @dependabot[bot] in #430
  • chore: pin Compose images to those of the chart by @uniqueg in #440
  • test: read config via get_config() by @uniqueg in #443
  • chore: remove unused .gitignore entries by @uniqueg in #445
  • ci: verify the release tag before its scripts by @uniqueg in #447
  • fix: read service infos without credentials by @uniqueg in #448
  • fix: end tracking of stuck CANCELING tasks by @uniqueg in #450
  • feat: find API paths of discovered TES instances by @uniqueg in #451
  • fix: limit the size of backend responses by @uniqueg in #452
  • fix: ignore .netrc in backend requests by @uniqueg in #454
  • build(deps): bump the python-dependencies group across 1 directory with 4 updates by @dependabot[bot] in #453
  • feat: limit tasks per user (opt-in) by @uniqueg in #456
  • feat: retain finished tasks for N days (opt-in) by @uniqueg in #457
  • fix: bound FULL task pages by size in bytes by @uniqueg in #458
  • build: bump FOCA to 0.22.0 by @uniqueg in #463
  • ci: label Dependabot PRs by @uniqueg in #461
  • test: cover validation_reasons turned off by @uniqueg in #470
  • build: pin the foca-plugins 0.3.0 release by @uniqueg in #471
  • build: prepare the 0.6.0 release by @uniqueg in #472

Full Changelog: v0.5.0...v0.6.0