Ticket 7893

Summary:	slurmctld core dump with acct_gather_infiniband/ofed
Product:	Slurm	Reporter:	Mathis Clayer <mathis.clayer>
Component:	Accounting	Assignee:	Gavin D. Howard <gavin>
Status:	RESOLVED FIXED	QA Contact:
Severity:	3 - Medium Impact
Priority:	---	CC:	kilian
Version:	19.05.2
Hardware:	Linux
OS:	Linux
See Also:	https://bugs.schedmd.com/show_bug.cgi?id=7940
Site:	Atos/Eviden Sites	Alineos Sites:	---
Atos/Eviden Sites:	Grenoble	Confidential Site:	---
Coreweave sites:	---	Cray Sites:	---
DS9 clusters:	---	HPCnow Sites:	---
HPE Sites:	---	IBM Sites:	---
NOAA SIte:	---	OCF Sites:	---
Recursion Pharma Sites:	---	SFW Sites:	---
SNIC sites:	---	Linux Distro:	---
Machine Name:		CLE Version:
Version Fixed:	19.05.4	Target Release:	---
DevPrio:	---	Emory-Cloud Sites:	---
Attachments:	slurm.conf

Description Mathis Clayer 2019-10-08 11:02:03 MDT

Hello,
With the option AcctGatherInfinibandType=acct_gather_infiniband/ofed, and without the file acct_gather.conf. I got a slurmctld coredump:

 slurmctld: debug2: No acct_gather.conf file (/usr/local/mathis/19.05/etc/acct_gather.conf)
slurmctld: debug2: before rc
slurmctld: debug2: after rc
slurmctld: debug2:  in s_p_pack_hashtbl
slurmctld: debug2:  in s_p_pack_hashtbl
slurmctld: error: parse_config.c:2190: s_p_pack_hashtbl(): Assertion (p) failed.
Aborted (core dumped)


If I create the file acct_gather.conf, I don't have a core dump anymore.

Comment 1 Gavin D. Howard 2019-10-08 13:48:00 MDT

Thank you for the report. I am looking into it.

Comment 2 Gavin D. Howard 2019-10-09 11:20:20 MDT

I am making progress, but I have a request: can you send me your slurm.conf?

Comment 3 Mathis Clayer 2019-10-10 07:02:34 MDT

Created attachment 11900 [details]
slurm.conf

Comment 5 Gavin D. Howard 2019-10-10 11:40:17 MDT

Good news! I have found the bug and have started the review process for a patch.

Comment 9 Gavin D. Howard 2019-10-16 12:21:02 MDT

*** Ticket 7940 has been marked as a duplicate of this ticket. ***

Comment 16 Gavin D. Howard 2019-10-25 16:23:20 MDT

This bug has been fixed, and the fix will be in 19.05.4.

I am closing this bug, but feel free to reopen if 19.05.4 has the same problem.