Project

General

Profile

Bug #20980

rgw-admin bi purge not removing bucket instance.

Added by Robin Johnson over 6 years ago. Updated over 6 years ago.

Status:
Closed
Priority:
Normal
Assignee:
-
Target version:
% Done:

0%

Source:
Tags:
Backport:
Regression:
No
Severity:
3 - minor
Reviewed:
Affected Versions:
ceph-qa-suite:
Pull request ID:
Crash signature (v1):
Crash signature (v2):

Description

This is an older install, migrated up from pre-Hammer to Jewel. Problem occurs on new buckets as well as older buckets.
It's not a multisite or federated deployment at all.

In a multisite deployment, I'd expect the bucket instance has to stay for sync purposes, but since it's not multisite, it should just be cleaned up. A cleanup tool like orphans-find for bucket instances & old bucket indices would also be very helpful here.

root@peon7929:/home/rjohnson# radosgw-admin metadata list bucket.instance |grep -e robjoh84-benjamin-test
    "robjoh84-benjamin-test:default.8061027.1",
    "robjoh84-benjamin-test:default.7801500.18876",
    "robjoh84-benjamin-test:default.8059830.1",
root@peon7929:/home/rjohnson# radosgw-admin bucket reshard --bucket robjoh84-benjamin-test --num-shards 34
*** NOTICE: operation will not remove old bucket index objects ***
***         these will need to be removed manually             ***
old bucket instance id: default.8061027.1
new bucket instance id: default.8061096.1
total entries: 15
root@peon7929:/home/rjohnson# radosgw-admin bi purge --bucket=robjoh84-benjamin-test --bucket-id=default.8061027.1
root@peon7929:/home/rjohnson# echo $?
0
root@peon7929:/home/rjohnson# radosgw-admin metadata list bucket.instance |grep -e robjoh84-benjamin-test
    "robjoh84-benjamin-test:default.8061096.1",
    "robjoh84-benjamin-test:default.8061027.1",
    "robjoh84-benjamin-test:default.7801500.18876",
    "robjoh84-benjamin-test:default.8059830.1",
root@peon7929:/home/rjohnson# radosgw-admin bi list --bucket=robjoh84-benjamin-test --bucket-id=default.8061027.1
ERROR: bi_list(): (2) No such file or directory

History

#1 Updated by Yehuda Sadeh over 6 years ago

Right. bi purge only removes the bucket index objects. There's 'radosgw-admin metadata rm bucket.instance:<bucket_id>' though. I think we should just have the resharding process purge and remove everything.

#2 Updated by Orit Wasserman over 6 years ago

  • Status changed from New to Closed

Also available in: Atom PDF