GitLab

This would work with whatever -e pyXY option tox is invoked with.

When a CeleryTask is deleted a postcommit hooks is trigerred to revoke the
tasks (and child tasks) and to remove the logs from the redis database.

This avoid filling the redis database with deleted task logs.

Suppose you have a test that create a task but not run it with run_all_tasks(),
then in a second test run_all_tasks() was running the task of the previous
test...
Fix this by ensuring the task_id (an unique hash generated by celery) is
reachable within the current test database.

This has been deprecated for a while, make this mandatory.

When a task create a new task (by calling start_async_task), _TEST_TASK was
reset during the loop on it, this was leading to a KeyError (in case of
multiple tasks) or in sub-tasks not being started.

Fix this by not overriding _TEST_TASK for each new cubicweb connection and by
consuming _TEST_TASK until there is no tasks left.

CELERY_ALWAYS_EAGER is already set in setUpClass

Just in case someone wants to keep the results the same even after running
run_all_tasks().

Under some mysterious circumstances, result.traceback appears to be bytes. So
decode it before trying to fire workflow transition with it as "comment"
value.

Use 0.4.0 which handle revoke event

When running tests in non EAGER mode, tasks workflow synchronization was done
in run_all_tasks(), this seemed to work because task execution was
asynchronously finished before call to celery-monitor.

Fix this by explicitely running celery-monitor in wait_async_task which wait
the task is actually finished.

When a cube inherit from CeleryTaskAdapter, celery-monitor was running
sync_task_state for each of them.

Replace this by adding a `sync_task_state()` function that first get or create
a CeleryTask, then adapt it to ICeleryTask and call `sync_state()` method.

This add a lot of test code to be maintained for a feature, provide an
alternative implementation than CeleryTaskAdapter, that actually has never been
used.

When a task is REVOKED, try to fire the 'fail' transition.

This require latest cset on cw-celerytask-helper [1] that add handling the REVOKED signal.

https://hg.logilab.org/review/cw-celerytask-helpers/rev/5ca553eb0f07

When a task is revoked, revoke child tasks as well.
Also add a `child_tasks` property on CeleryTaskAdapter to recurse over child
tasks.

In tests set the name of the worker to te@st so it's easier to get the revoke
task list from the worker.

When run_all_tasks() is called multiple times in the test, tasks where run
twice. Ensure _TEST_TASKS is empty after run_all_tasks().

"running" is a state, not a transition.

Closes #17057660.

This test the fix in cw-celerytask-helpers: https://www.cubicweb.org/patch/17056309

Previous cset was rebased and published.
Also remove the '#egg=cw-celerytask-helpers' since it's only used for git or hg
urls.

PROGRESS is a custom state used to handle progress bar in views.
When we receive PROGRESS, fire transition 'running' (if possible).

Convention is to return a dict with a key "celerytask_subtasks".
So we can remove a "except Exception" that could hide errors or future bugs.

Set CW_CONNECTIONS_POOL_SIZE to 1 (by default) to only use one connection to
the database (no more is needed since it's single threaded).

Setting the parameter in the "config" object could not work since CubicWeb
reload it from config files when loading cubes.