Bug 2222663 - potential memory leak in sssd_autofs
Summary: potential memory leak in sssd_autofs
Keywords:
Status: NEW
Alias: None
Product: Red Hat Enterprise Linux 7
Classification: Red Hat
Component: sssd
Version: 7.9
Hardware: All
OS: Linux
unspecified
medium
Target Milestone: rc
: ---
Assignee: Pavel Březina
QA Contact: sssd-qe
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2023-07-13 11:44 UTC by Renaud Marigny
Modified: 2023-07-21 12:15 UTC (History)
2 users (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
Environment:
Last Closed:
Target Upstream Version:
Embargoed:


Attachments (Terms of Use)


Links
System ID Private Priority Status Summary Last Updated
Red Hat Issue Tracker RHELPLAN-162274 0 None None None 2023-07-13 11:45:54 UTC
Red Hat Issue Tracker SSSD-6479 0 None None None 2023-07-21 12:15:40 UTC

Description Renaud Marigny 2023-07-13 11:44:39 UTC
Description of problem:

We have a customer experimenting a memory growth using sssd_autofs. This generally ends in OOM condition and the process sssd_autofs being killed

Doing 100 "automount -m" runs the memory usage increases by 0.7GB, so the memory increment is just over 7MB per execution.

A large automount map is used in IPA, "automount -m | wc -l" returns 2255 lines

A valgrind capture has been performed on sssd_autofs. This reveals no "definitely lost" bytes but quite a lot of "possibly lost" like here :

valgrind.32064.log:==00:00:01:53.245 32064==      possibly lost: 6,313,884 bytes in 38,925 blocks

Version-Release number of selected component (if applicable):

RHEL 7.9

How reproducible:

Steps to Reproduce:

Not reproduced in-house for now

Actual results:


Expected results:

sssd_autofs process size should remain stable over time

Additional info:

Comment 4 Pavel Březina 2023-07-21 12:10:22 UTC
Thank you. I can reproduce the issue on RHEL-7 (sssd-1.16). It was fixed as a side effect of refactoring autofs code in RHEL-8 so it does not affect RHEL-8 nor RHEL-9.


Note You need to log in before you can comment on or make changes to this bug.