NFS Erratic Behavior - BUG?

[Kei]

Hello,
I'm having a weird problem with all my VM's residing on a remote NFS share. The already running VM's dont have a single problem, infact I can see that the last modified date increments on the NAS server, so that means the connectivity is good. However, if I stop and try to restart any of the VM's, I get this error:

TASK ERROR: storage 'HEZE_VM' is not online

This is very weird also because I can navigate on the mount point and I can correctly view the remote files with no problem:

192.168.60.72:/mnt/Data/Proxmox/BKP  323G  7,0G  316G   3% /mnt/pve/HEZE_BKP
192.168.60.72:/mnt/Data/Proxmox/ISO   54G  251M   54G   1% /mnt/pve/HEZE_ISO
192.168.60.72:/mnt/Data/Proxmox/VM   162G   11G  151G   7% /mnt/pve/HEZE_VM

root@pve1:~# ls -la /mnt/pve/HEZE_VM/images/110/
total 4529902
drwxr-xr-x 2 root root          3 nov 19 01:19 .
drwxr-xr-x 9 root root          9 nov 19 01:30 ..
-rw-r----- 1 root root 8589934592 nov 19 14:27 vm-110-disk-1.raw

This seems to be affecting both nodes of my cluster.
What should I try?

Thanks in advance

 

[fabian]

you can check whether the command "showmount --no-headers --exports SERVER" works, this is what we use for checking whether an NFS storage is online. if it doesn't work, maybe you have some firewall that blocks this request?

 

[Kei]

Thank you Fabian, I might have it figured out. Apparently it's not proxmox fault.
I realized this when I found out I was prevented to initiate new connections form different clients too, even thought all the ongoing connections were fine. The problem seemed to be caused by the NAS, (I use FreeNAS 9.10) that apparently starts dropping TCP connections (netstat -np tcp), only associated with NFS, when a particular VM on Proxmox is started and connected to the network via two NICs using PCI passthru. I know, it's weird. I managed to solve this problem only by deleting and recreating the interface dedicated to the NFS traffic in Freenas. I hope this is informative for somebody running in this kind of problem.