Actions
Feature #50966
openrgw: bnucket index resharding with non-blocking write I/Os
Status:
New
Priority:
Normal
Assignee:
-
Target version:
-
% Done:
0%
Source:
Tags:
Backport:
Reviewed:
Affected Versions:
Pull request ID:
Description
I have a Ceph cluster (octopus) for RGW. The number of objests grows up periodically.
For example, my logging system (Loki) stores its data to RGW.
Of course, I'll set suffiliently large number of shards on bucket creation, if possible.
However, there are many buckets in which the estimation of this value is hard.
When I ran resharding for a backet which contains about 20 millions of objects,
it takes over 10 minutes. It means that I can't write something to this bucket
over 10 minutes. It's unacceptable for me.
FYI, here is an issue which are caused by two problems: resharding blocks write I/Os; resharding takes too long time.
https://tracker.ceph.com/issues/39485
No data to display
Actions