Hi Kotresh,
The issue is fixed for now I followed the steps below.
I have an unmounted kernel client and restarted mds service which brought
back mds to normal. But even after this "1 MDSs behind on trimming issue"
didn't solve I waited for about 20 - 30 mins which automatically fixed the
The MDS requests the clients to release caps to trim caches when there is
cache pressure or it
might proactively request the client to release caps in some cases. But the
client is failing to release the
caps soon enough in your case.
Few questions:
1. Have you tuned MDS cache configurations? If
