Bug #57951
openrgw: lc: lc for a single large bucket can run too long
0%
Description
If this happens, other lc hosts/threads can attempt to process the same bucket, which inflates overhead without any corresponding benefit. In practice, this may be likely only during test runs with a debug interval (which, in turn, should never be used in production), but we should prevent the behavior anyway.
Updated by Casey Bodley over 1 year ago
- Status changed from In Progress to Fix Under Review
- Tags set to lifecycle
- Pull request ID set to 48683
Updated by J. Eric Ivancich about 1 year ago
- Status changed from Fix Under Review to Pending Backport
Updated by Backport Bot about 1 year ago
- Copied to Backport #58787: pacific: rgw: lc: lc for a single large bucket can run too long added
Updated by Backport Bot about 1 year ago
- Copied to Backport #58788: quincy: rgw: lc: lc for a single large bucket can run too long added
Updated by Backport Bot about 1 year ago
- Tags changed from lifecycle to lifecycle backport_processed
Updated by Cory Snyder 11 months ago
We seem to have encountered this scenario many times on production Pacific clusters. At least, we have a lot of error logs that indicate LC is processing the same bucket on multiple hosts concurrently over the period of multiple days. Can I assist with backports on this one, Matt?
Updated by Matt Benjamin 11 months ago
Hi Cory,
Sure, I'd be delighted to have you do that, and will review and assist if needed.
Matt
Updated by Cory Snyder 7 months ago
For what it's worth, I'll note that we've commonly seen lifecycle rules be used to delete everything in a bucket when a client wishes to empty it so that the bucket itself may be deleted. For buckets with tens or hundreds of millions of objects, and LC rules that are designed to immediately expire every object in the bucket, the processing can easily take multiple days.