]> git-server-git.apps.pok.os.sepia.ceph.com Git - ceph.git/commit
cls/rgw: keep issuing bilog trim ops after reset 29326/head
authorCasey Bodley <cbodley@redhat.com>
Wed, 5 Jun 2019 21:07:47 +0000 (17:07 -0400)
committerNathan Cutler <ncutler@suse.com>
Thu, 25 Jul 2019 16:13:02 +0000 (18:13 +0200)
commit4c4def2fc34e71b002b5809410b9ccb44d54750e
tree3cd576690b6a473a7e6e62abac51cb57ec96e59c
parent85fa207048f84d9c6125070861de8ae3d02b3c7b
cls/rgw: keep issuing bilog trim ops after reset

The CLSRGWIssueBILogTrim class is intended to keep submitting
cls_log_trim operations on each bucket index shard until they return
-ENODATA to signal there's no more to trim.

However, if the while loop in CLSRGWConcurrentIO::operator() gets to the
end of the shard list and collects all of its completions,
wait_for_completions() will return false and exit the loop. This is
always the case when num_shards = 1.

Fixes: http://tracker.ceph.com/issues/40187
Signed-off-by: Casey Bodley <cbodley@redhat.com>
(cherry picked from commit a6ea8433fe9eb032180368bc3d36a4232dd6d08b)
src/cls/rgw/cls_rgw_client.h