After seeing this thread again, and still not resolved, Iām realizing this is Proxmox (Debian 12). Have you gone through the steps to fix the portblock resource agent on all nodes? Itās documented in this blog post under āPortblock Resource Agent Workaroundā. This is necessary for both NFS and iSCSI.
Long story short, Debian 12 ships a version of iptables with a regression that interacts with the portblock resource agent version they also ship. The newer resource agents are able to workaround the behavior of the specific iptables version, and later versions of iptables reverted the behavior entirely.
The fix:
Copy the portblock resource agent from upstream to each node:
wget -O /usr/lib/ocf/resource.d/heartbeat/portblock \
https://raw.githubusercontent.com/ClusterLabs/resource-agents/main/heartbeat/portblock
If you have an existing export, Iād just delete and recreate it and see if that resolves the behavior.
Without this fix the resource would āflapā between hosts in my cluster, seemingly juggling where NFS was active, but running linstor-gateway nfs list showed normal output before I caught on to the flapping behavior.