nsinnovations/erpnext
2
0
Fork 0
mirror of https://github.com/frappe/erpnext.git synced 2026-05-01 04:28:27 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix: use RecoverableErrors isinstance check for repost timeout status (backport #54543) (#54649)

Browse Source 
fix: use RecoverableErrors isinstance check for repost timeout status

When a Repost Item Valuation job is killed by an RQ worker timeout
(JobTimeoutException raised via SIGALRM), the existing status detection
relied solely on traceback string matching for 'timeout' or 'Deadlock'.

This is unreliable because SIGALRM can interrupt a C-extension call
(e.g. inside pypika's copy.copy()) before Python records the exception
in the traceback. In that case the traceback shows only the interrupted
frame -- not JobTimeoutException -- so the job is permanently marked
'Failed' instead of 'In Progress', preventing the scheduler from
automatically retrying it.

RecoverableErrors = (JobTimeoutException, QueryDeadlockError,
QueryTimeoutError) is already defined at the top of this file and is
already used further down in the same except block to suppress email
notifications. Extend its use to also guard the status decision.

The traceback string fallback is kept as a secondary check for
forward compatibility with other timeout signals.

Fixes: jobs permanently stuck as 'Failed' after RQ worker timeout,
requiring manual re-queue to resume reposting.

(cherry picked from commit a49e2de866)

Co-authored-by: Assem Bahnasy <bahnasyassem@gmail.com>
This commit is contained in:
mergify[bot]
2026-04-29 12:02:04 +00:00
committed by GitHub
parent 86cf256358
commit b3001595ab
1 changed files with 9 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
erpnext/stock/doctype/repost_item_valuation/repost_item_valuation.py
View File

@@ -411,8 +411,15 @@ def repost(doc):
message = message.get("message")
status = "Failed"
# If failed because of timeout, set status to In Progress
if traceback and ("timeout" in traceback.lower() or "Deadlock found" in traceback):
# If failed because of a recoverable error (timeout, deadlock), set status to In Progress
# so the scheduler automatically retries instead of leaving it permanently failed.
# NOTE: isinstance check comes first because the traceback string matching is unreliable
# when SIGALRM kills the process mid-C-extension (JobTimeoutException may not appear
# in the traceback if the exception handler itself was interrupted).
traceback_lower = traceback.lower() if traceback else ""
if isinstance(e, RecoverableErrors) or (
traceback_lower and ("timeout" in traceback_lower or "deadlock found" in traceback_lower)
):
status = "In Progress"
if traceback: