Proxmox and NFS with DRBD

[taich]

Hi,

I run a Proxmox Cluster with 3 nodes and a Pacemaker Cluster with NFS on 2 nodes.
I use the NFS Cluster as a HA storage.

All that worked fine for a while. I could switch from NFS node1 to node2 without problems.
Running VM's on Proxmox did not recocnize the change.

But since 2 or 3 months all VM's have disks monted read only after every switch of the NFS storage.

Why cant the VM's survive a storage cluster switch?

 

[bbgeek17]

There should be information in dmesg/journal as to why Kernel marks NFS r/o. Most likely the transition takes longer than default timeout. If that is the case, one solution is to increase the timeout. Another approach would be to figure out why the transition takes longer than default timeout. Or you can switch to hard mount.

---

Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

 

[taich]

I could not find where to increase increase timeout, or hard mount. Where does this configuration happen?

 

[bbgeek17]

You didnt say whether you use NFS directly in VM or as storage object in Proxmox. Overall it doesnt matter because the settings are the same just in different location.

https://linux.die.net/man/5/nfs
timeo=n
soft / hard

https://pve.proxmox.com/wiki/Storage:_NFS
options vers=3,soft

---

Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

 

[taich]

Oh sorry, I use it as a proxmox storage, where VM's have their "harddisk".

Thank you for the hint.