Rakshith R

@akash123-eng network issues/instability may cause ceph client to be blocklisted. That would be the first place to check.

> Yes @Rakshith-R it's may be due to network issue but as I mentioned in the issue blocklisted client ip is pod ip of mgr pod scheduled on other node...

> @Rakshith-R Yes but whenever we face this issue ceph status is all healthy and fine. And also this happens on any random node. So I just wanted to know...

@akash123-eng There is an option to recover_session for cephfs kernel mounter. I haven't tested it anytime but it may help. https://docs.ceph.com/en/quincy/man/8/mount.ceph/#basic you can add it here https://github.com/rook/rook/blob/4817a0442cbe96dee0d60b52acd06b1fae204ff2/Documentation/CRDs/Cluster/ceph-cluster-crd.md#csi-driver-options

> please provide cephfs plugin logs from the node where you have the problem, make sure you set `CSI_LOG_LEVEL` to `5` ad CSI_SIDECAR_LOG_LEVEL`to`5` as well This isn't provided yet. Provide...

https://github.com/actions/runner/issues/2676

> As we already know currently, a csiaddons node object is created, we create the connections and keep it until the addons node object is deleted. there could be advantages/disadvantages...

>Disadvantages >- More number of idle connections in in-memory (if there are no addons operations) >- More resources utilization in the controller pod >- More network calls as connection keep-alive...

> There is no real need to connect to all CSIAddonsNodes on startup of the controller, and only connect on-demand. This is needed to fetch capabilities and store it. >...

> @zerotens, Thanks for the PR 😃 I had already started working on this issue and was testing the same, and have opened a PR now #4661 > > >...