Bug #9887
closedlast_scrub_stamp changes twice after scrub is started and confuses test case
0%
Description
I'm trying to add a yaml that uses the test case tasks/repair_test.py. For some reason the last_scrub_stamp is changing twice. The python routine do_pg_scrub() is supposed to wait for the scrub to finish, but because of the last_scrub_stamp changing it returns assuming that the scrub it initiated has finished.
2014-10-24 21:14:31,559.559 INFO:tasks.ceph.ceph_manager:stamp 2014-10-24 21:14:27.070848
Start scrub
2014-10-24 21:14:35,832.832 INFO:tasks.ceph.ceph_manager:final stamp 2014-10-24 21:14:27.212924 still active+clean
cep pg dump 11 seconds later
active+clean+inconsistent 2014-10-24 21:14:38.652441
This change of the last_scrub_stamp makes the test believe that the scrubbing has completed. By the time I look at the machine after the test pauses at that point, the last_scrub_stamp has advanced again and the state is as expected by the test case. I don't understand why the last_scrub_stamp is changing twice.
Updated by David Zafman over 9 years ago
- Status changed from New to Fix Under Review
Reviewing of wip-9887 in progress. Haven't tested with old/new OSDs. Related to wip-dzaddscrub in ceph-qa-suite repo which can go in after this does.
Updated by Tamilarasi muthamizhan over 9 years ago
- Assignee changed from David Zafman to Anonymous
Updated by Samuel Just over 9 years ago
- Status changed from Fix Under Review to Resolved