[18.0][FIX] Don't try to enqueue dependent jobs when job failed

[thomaspaulb]

Running (on 16.0) into:

2026-01-23 13:45:15,992 351151 ERROR custdatabase odoo.sql_db: bad query: UPDATE "stock_quant" SET "reserved_quantity" = '45212.0000', "write_date" = '2026-01-23 13:45:15.153636', "write_uid" = 1 WHERE id IN (1793638)
ERROR: could not serialize access due to concurrent update
2026-01-23 13:45:15,994 351151 DEBUG custdatabase odoo.addons.queue_job.controllers.main: <Job 7f20f326-3383-4c42-9cdf-7126dbd89743, priority:2> OperationalError, postponed 2026-01-23 13:45:16,000 351151 DEBUG custdatabase odoo.addons.queue_job.controllers.main: <Job 7f20f326-3383-4c42-9cdf-7126dbd89743, priority:2> postponed 2026-01-23 13:45:16,001 351151 DEBUG custdatabase odoo.addons.queue_job.controllers.main: <Job 7f20f326-3383-4c42-9cdf-7126dbd89743, priority:2> enqueue depends started File "/data/home/odoo/odoo/auto/addons/muk_rest/models/ir_http.py", line 160, in _dispatch
response = super()._dispatch(endpoint)
File "/data/home/odoo/odoo/.venv/lib/python3.10/site-packages/odoo/addons/base/models/ir_http.py", line 154, in _dispatch
result = endpoint(**request.params)
File "/data/home/odoo/odoo/.venv/lib/python3.10/site-packages/odoo/http.py", line 761, in route_wrapper
result = endpoint(self, *args, **params_ok)
File "/data/home/odoo/odoo/auto/addons/queue_job/controllers/main.py", line 200, in runjob
self._runjob(env, job)
File "/data/home/odoo/odoo/auto/addons/queue_job/controllers/main.py", line 169, in _runjob
cls._enqueue_dependent_jobs(env, job)
File "/data/home/odoo/odoo/auto/addons/queue_job/controllers/main.py", line 86, in _enqueue_dependent_jobs
with job.env.cr.savepoint():
File "/data/home/odoo/odoo/.venv/lib/python3.10/site-packages/odoo/sql_db.py", line 157, in savepoint
return _FlushingSavepoint(self)
File "/data/home/odoo/odoo/.venv/lib/python3.10/site-packages/odoo/sql_db.py", line 101, in __init__
super().__init__(cr)
File "/data/home/odoo/odoo/.venv/lib/python3.10/site-packages/odoo/sql_db.py", line 76, in __init__
cr.execute(SQL('SAVEPOINT {}').format(self._name))
File "/data/home/odoo/odoo/.venv/lib/python3.10/site-packages/odoo/sql_db.py", line 327, in execute
_logger.error("bad query: %s\nERROR: %s", tools.ustr(self._obj.query or query), e)
File "/data/home/odoo/odoo/.venv/lib/python3.10/site-packages/odoo/sql_db.py", line 485, in __getattr__
raise psycopg2.InterfaceError("Cursor already closed")
psycopg2.InterfaceError: Cursor already closed

This is because job.env.cr is closed at that point, but it's still being used here which may fail. Not the end of the world if it fails, but clutter for the logfile.

Not sure if this is the best solution. I thought about opening a new fresh job cursor, but then again, why would dependent jobs need to be enqueued if the original job failed?

[OCA-git-bot]

Hi @guewen, @sbidoul,
some modules you are maintaining are being modified, check this out!

[sbidoul]

Do you hit this after a RetryableJobError? Maybe we need a returnhere because all other cases raise?

But see also #884 which I just opened.

[thomaspaulb]

Yes, after a retryablejoberror.

[sbidoul]

This should be covered by #900, so I'm going to close this one. Thanks for point this out!