Bug #44268
openmultisite/lc: lc doesn't run in the slave
0%
Description
From the master, if I set a lilfecycle policy for a bucket, the lc thread of the slave cluster doesn't process that bucket.
But from the slave, if I set it, the slave lc starts to work.
Updated by Ilsoo Byun about 4 years ago
I created a related PR(https://tracker.ceph.com/issues/44268).
Updated by Casey Bodley about 4 years ago
- Status changed from New to In Progress
- Tags set to multisite lifecycle
- Backport set to nautilus octopus
Updated by Ilsoo Byun about 4 years ago
I made a PR to the master branch.
https://github.com/ceph/ceph/pull/34553
Updated by Matt Benjamin almost 4 years ago
Hi @Ilsoo Byun I think this is a valid improvement, but I think we need to handle the special case where we are syncing to an archive zone. In that case, the current consensus is that ordinary lifecycle policy must not take effect on the archive zone--otherwise, data would tend to be eliminated there precisely when it was expired on the active sites, preventing recovery from the archive. We think it does make sense to enable explicit policy for data reduction/expiration in archive zone, but we thought a) such policy should need to be applied explicitly in the archive zone; and b) possibly needs to be introduced as an extension to the lifecycle grammar so that ordinary policies can't be applied by mistake.
Updated by Nathan Cutler almost 4 years ago
- Status changed from In Progress to Fix Under Review
- Pull request ID set to 34553
Updated by Casey Bodley over 2 years ago
- Assignee set to Matt Benjamin
- Backport changed from nautilus octopus to octopus pacific
Updated by Casey Bodley over 2 years ago
- Related to Bug #43805: bucket lifecycle breaks down when master-zone changed or period gets updated added
Updated by Casey Bodley almost 2 years ago
- Related to Bug #55487: bucket lifecycle for aborting old multipart uploads does not work on non-master multisite zone buckets added