p-alik / perl-Gearman / 26718993435

86%

master: 86%

Build Type

push

github

Committed by p-alik

Commit Message

Worker: don't die on a stray error packet during grab_job

work_complete/work_fail/work_status are sent fire-and-forget. When a
job server has already timed out a job server-side, it rejects the
worker's later work result with an asynchronous error packet
(JOB_NOT_FOUND), which only surfaces on the next grab_job read.

Previously the worker die()d on that packet, killing the whole worker
process and hanging any client waiting on a subsequent job. t/14-sleep.t
reproduces this against gearmand 1.1.20 (it intentionally runs jobs past
a CAN_DO_TIMEOUT), which is why CI hung.

Treat a stray error packet during grab_job as non-fatal: log it under
debug and move on to the next job server.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Signed-off-by: Alexei Pastuchov <info@maximka.de>

Coverage Stats

0 of 5 new or added lines in 1 file covered. (0.0%)

99 existing lines in 6 files now uncovered.

957 of 1115 relevant lines covered (85.83%)

32.25 hits per line