Ceph » rgw

https://github.com/ceph/ceph/pull/18922

Aleksei Gutikov wrote:

https://github.com/ceph/ceph/pull/18922

Sorry, this is not the fix, mixed up with http://tracker.ceph.com/issues/22122

We probably need to remove the stats entry for the old bucket instance from the user stats.

need also to reset user stats

https://github.com/ceph/ceph/pull/19253

@Orit Wasserman Wasserman， the pr https://github.com/ceph/ceph/pull/19253 has introduced a new bug, after this pr, execute the command `./bin/radosgw-admin bucket reshard --num-shards=10 --bucket=xxx` and the bucket xxx is missing from get service list，here is a test:

1. s3cmd mb s3://testbucket
   Bucket 's3://testbucket/' created
2. s3cmd ls
   2018-01-11 06:48 s3://testbucket
3. ./bin/radosgw-admin bucket reshard --bucket=testbucket --num-shards=10
   2018-01-11 14:49:00.964 7fedf81ab3c0 -1 WARNING: all dangerous and experimental features are enabled.
   2018-01-11 14:49:00.964 7fedf81ab3c0 -1 WARNING: all dangerous and experimental features are enabled.
   2018-01-11 14:49:00.984 7fedf81ab3c0 -1 WARNING: all dangerous and experimental features are enabled.
   2018-01-11 14:49:01.424 7fedf81ab3c0 0 WARNING: detected a version of libcurl which contains a bug in curl_multi_wait(). enabling a workaround that may degrade performance slightly.
   • NOTICE: operation will not remove old bucket index objects ***
   • these will need to be removed manually ***
     tenant:
     bucket name: testbucket
     old bucket instance id: 071ae4cf-ffb2-4a82-8b7a-ca1ebd5e204f.4153.1
     new bucket instance id: 071ae4cf-ffb2-4a82-8b7a-ca1ebd5e204f.4148.1
     total entries: 0
4. s3cmd ls #

the testbucket is missing.

And I have tested even before this commit, the user stats not increased after bucket reshard.

Maybe we need to revert the commit.

I will check it

@Zongyou Yao thanks for catching this.
I revert the pr and I will try a different fix.

new fix:
https://github.com/ceph/ceph/pull/20038

merged luminous backport: https://github.com/ceph/ceph/pull/20327