Actions
Bug #17998
closedceph-osd oom killed during rados bench teuthology run
Status:
Closed
Priority:
Urgent
Assignee:
-
Category:
-
Target version:
-
% Done:
0%
Source:
Tags:
Backport:
Regression:
No
Severity:
3 - minor
Reviewed:
Affected Versions:
ceph-qa-suite:
Pull request ID:
Crash signature (v1):
Crash signature (v2):
Description
Notably, this was btrfs. Looks like 2 osd processes were running 2-3 gb and another was 4-5. That's not necessarily a leak given this test involved pgs being aggressively split way above what would normally be reasonable.
samuelj@teuthology:/a/samuelj-2016-11-22_03:24:33-rados-wip-sam-testing---basic-smithi/568012/remote
You can see the oom kills with zcat remote/*/syslog/kern.log.gz
I bet this is easy to reproduce by running that same task 5 times.
Actions