Reconnect NFS after it's disconnected (without rebooting)

M

Michael Mansour

Member
Oct 29, 2018
4
0
6
53
Hi,

I have a 3 node Proxmox 6.3 cluster, and a Proxmox backup server.

From the Proxmox backup server I export 3 NFS shares to the 3 node Proxmox cluster.

When I patch the Proxmox Backup server ready for a reboot, I:

  1. go to the 3 node cluster Datacenter -> Storage
  2. Disable the 3 NFS shares and the PBS share
  3. patch the Proxmox backup server
  4. Reboot the Proxmox backup server
When Proxmox backup server is completely up, I:

  1. go to the 3 node cluster Datacenter -> Storage
  2. Re-enable the 3 NFS shares and the PBS share
The PBS share comes up (and is fact enabled and online in the left storage list), the 3 NFS shares don't, they remain offline.

A sequenced reboot of each Proxmox cluster node will bring the NFS shares back online.

I'm wondering how I can overcome this problem ie. I don't want to reboot the cluster nodes?

I run:

pvesm status

while the NFS shared are offline, and it does show the daemon is unable to communicate with the NFS shares.

I just don't want to reboot the Proxmox cluster sequentially every time I want to reboot the Proxmox backup server.

Any ideas why this happens and how I can work around it, or not be forced to reboot each physical host after rebooting the Proxmox backup server ?

PS. as a side note, I also have a single Proxmox node outside of the 3 node Proxmox cluster which is my test Proxmox node, it's also mapped to the 3 NFS shares and PBS share, but it does NOT disconnect the NFS shares after the Proxmox Backup server is reboot, the NFS shares stay online. For that server, the disable/re-enable of the NFS shares isn't done in "Datacenter - Storage" on that singular test server.

Thank you.

Michael.
 
Last edited:
Hi Alwin, thanks for your reply.

I remember I did this first time and it caused major problems for the cluster nodes ie. they all lost connections to one another and couldn't see each other, I had to forcefully reboot each node one by one to get the cluster back up and running.

The process that time was:

  1. 3 node proxmox cluster, 1 proxmox backup server
  2. reboot the backup server
  3. all 3 nodes lost connectivity to one another aka the proxmox cluster failed
  4. had to manually reboot node 1 then node 2 then node 3, proxmox cluster back up
So for point 3:
  • node 1 sees it's VM but disconnections to node 2 and 3
  • node 2 sees it's VM but disconnections to node 1 and 3
  • node 3 sees it's VM but disconnections to node 1 and 2
I can try it again and advise further, but I avoided that way since the above happened.
 
Michael Mansour said:
So for point 3:
Click to expand...
Cluster communication shouldn't be interrupted by an unavailable storage. The GUI though may see the nodes disconnected, the pvestatd might hang for the duration of the storage outage. Best check with pvecem status, while the backup server reboots.
 
You must log in or register to reply here.
Top Bottom