osd/osd-rep-recov-eio.sh: TEST_rados_repair_warning: return 1#37483
osd/osd-rep-recov-eio.sh: TEST_rados_repair_warning: return 1#37483neha-ojha merged 2 commits intoceph:masterfrom
Conversation
Signed-off-by: David Zafman <dzafman@redhat.com>
|
Testing passed with other failures |
| rados_get $dir $poolname ${objbase}-$i || return 1 | ||
| done | ||
|
|
||
| wait_for_clean |
There was a problem hiding this comment.
has the test been failing for lack of clean just at this stage or everywhere else?
There was a problem hiding this comment.
After a read that needs a repair, recovery is started. I never noticed on my build machine that the query could happen before the recovery is done, but teuthology did have this race. I did a sleep then Brad asked if there was some other way to tell of the repair is finished. So I realized that wait_for_clean() would do it. The only other possible race, is if recovery was async to the read repair. So if after the read the PG was still active+clean we would have a problem.
There was a problem hiding this comment.
I've just seen the test fail here so wondering whether we need to add wait_for_clean in all the other places or not, certainly doesn't hurt
Fixes: https://tracker.ceph.com/issues/46405 Signed-off-by: David Zafman <dzafman@redhat.com>
|
@dzafman is this ready to merge? |
|
@neha-ojha Yes |
Checklist
Show available Jenkins commands
jenkins retest this pleasejenkins test classic perfjenkins test crimson perfjenkins test signedjenkins test make checkjenkins test make check arm64jenkins test submodulesjenkins test dashboardjenkins test apijenkins test docsjenkins render docsjenkins test ceph-volume alljenkins test ceph-volume tox