Actions
Fix #65408
openBug #65314: valgrind error: Leak_PossiblyLost posix_memalign UnknownInlinedFun ceph::buffer::v15_2_0::list::refill_append_space(unsigned int)
qa: under valgrind, restart valgrind/mds when MDS exits with 0
Status:
Fix Under Review
Priority:
High
Assignee:
Category:
Code Hygiene
Target version:
% Done:
0%
Source:
Q/A
Tags:
Backport:
squid,reef,quincy
Reviewed:
Affected Versions:
ceph-qa-suite:
Description
Instead of issuing a re-execve
when respawning, the MDS should exit status 0 when a development config is turned on (used in fs:valgrind). This allows the normal atexit routines and destructors to be called without valgrind reporting some spurious errors in its logs.
Updated by Patrick Donnelly 22 days ago
(Trying to see if redmine adds Venky to the "Watchers" list)
Updated by Patrick Donnelly 22 days ago
- Status changed from In Progress to Fix Under Review
- Pull request ID set to 56825
So, the mds_valgrind_exit already exists and is turned on. The original problem in #65314 wasn't caused by a failover either.
Actions