New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
rgw: fix rgw.none statistics #43103
rgw: fix rgw.none statistics #43103
Conversation
Only unaccount for objects being removed if they exist. Also, clean up some formatting. Signed-off-by: J. Eric Ivancich <ivancich@redhat.com>
6524a4a
to
95b662f
Compare
@cbodley , @mattbenjamin : Could I get your eyes on this? Thanks! |
sorry @ivancich! this change looks correct to me in theory. but without a reproducer, or any test coverage of these bucket stats, it's hard to be confident that it solves the issue without any regression i also don't want to burden you with the impossible task of proving that stats are always consistent. can you think of some minimal test coverage that could help give us confidence here? or do you feel that this change is trivial enough to be confident without test coverage here? |
@ivancich because the bucket stats are managed internally by cls_rgw for each index shard, i think an approach similar to ceph_test_cls_rgw (which only tests the cls_rgw_client api against a single rados object) is the easiest way to get test coverage for stats. in #43843 i started on a new test that throws a bunch of random create/delete operations at the index, and tracks the expected bucket stats in memory for comparison |
i updated #43843 to model multipart uploads, and was a) able to reproduce these negative rgw.none stats from canceled multiparts, and b) verify that this pr resolved the issue |
Only unaccount for objects being removed if they exist.
Also, clean up some formatting.
Signed-off-by: J. Eric Ivancich ivancich@redhat.com
Fixes: https://tracker.ceph.com/issues/51560