From 3bb188839338b44e41be21dde7c5261442593c6b Mon Sep 17 00:00:00 2001
From: Moe Jette <jette1@llnl.gov>
Date: Mon, 3 Nov 2003 20:48:28 +0000
Subject: [PATCH] Log a shared memory anomaly using debug() rather than error.
The event is the natural consequence of a job ending immediately and before
the shared memory state information is completely established. (gnats:321)
---
src/slurmd/mgr.c | 10 ++++++++--
1 file changed, 8 insertions(+), 2 deletions(-)
diff --git a/src/slurmd/mgr.c b/src/slurmd/mgr.c
index aa9f918d128..d62153cbcf0 100644
--- a/src/slurmd/mgr.c
+++ b/src/slurmd/mgr.c
@@ -555,13 +555,19 @@ _create_job_session(slurmd_job_t *job)
return ESLURMD_FORK_FAILED;
}
+ /*
+ * If the created job terminates immediately, the shared memory
+ * record can be purged before we canset the mpid and sid below.
+ * This does not truly indicate an error condition, but a rare
+ * timing anomaly. Thus we log the event using debug()
+ */
job->jmgr_pid = getpid();
if (shm_update_step_mpid(job->jobid, job->stepid, getpid()) < 0)
- error("shm_update_step_mpid: %m");
+ debug("shm_update_step_mpid: %m");
job->smgr_pid = spid;
if (shm_update_step_sid(job->jobid, job->stepid, spid) < 0)
- error("shm_update_step_sid: %m");
+ debug("shm_update_step_sid: %m");
/*
* Read information from session manager slurmd
--
GitLab