Hi Drew,
On Thursday, March 4th, 2021 at 15:18, Drew Weaver <drew.weaver(a)thenap.com> wrote:
Howdy, the dashboard on our cluster keeps showing
LARGE_OMAP_OBJECTS.
I went through this document
https://www.suse.com/support/kb/doc/?id=000019698
I've found that we have a total of 5 buckets, each one is owned by a different user.
Do you have dynamic sharding enabled? If so, hitting the large OMAP object threshold is a
bit suspicious, as resharding should keep each shard below the threshold.
Did you look into your logs to find out which object is affected, and it's the number
of keys or its size triggering the warning (you can grep for "Large omap object
found" in /var/log/ceph/ceph.log)?
From the name of the object, you can figure out which bucket is affected, and you can look
for it in the default.rgw.buckets.index pool and see if the index keys are evenly
distributed among the shards.
--
Ben