Disconnect hanging/invalid nfs-server

[ggschmidt]

Hello all,
new to proxmox I got some trouble with nfs for external backup.

After problems with an nfs-server I could delete it in storage-area of the web-administration and newly add under a different name (same server).
This site is ok.

But browsing in the procmox file system (by ssh-shell) it hangs when browsing to /mnt/pve or display "df" (seems nfs still tries to connect).
start/stop etc service nfs-common does not affect anything, no entry in fstab as usal.

Any Idea how to finally delete and stop nfs connections?

thx
gs

 

[dendi]

I just opened the forum to write a new post and I found this.
I'm trying to remove a nfs storage, moved all VMs disks to a new storage, disabled storage but I see it in df.
I tried to umount it but I got "resource busy".
Double cheked all VMs, there are only some unused disks on the old storage, I don't want to remove because I want a backup on the old storage, I just want to shut down it in a safe way.
PVE 3.0.20 with 3 node cluster

So ggschmidt I suggest you to check if some process VM has disks on the old storage and try to unmount the nfs via ssh and wait with me for some proxmox expert

 

[dendi]

I solved my problem, you can find wich processes are using the storage with the command lsof
example:
lsof | grep name-of-the-mount
then you get the pid on the first column and you can type:
ps auxw | grep pid
and find the VM,remove the disk or cdrom in that storage and reboot it.

I notice that if I disable the storage without deleting it, it is still mounted and I have to unmount it manually, now with no errors, is it ok?

 

[ggschmidt]

Hello Dendi,
your problem has nothing to do with mine. It only killed my thread :-(

So far, I found the thing, I looked for myself. The line I looked for in fstab, is not there but is in /etc/pve/storage.cfg.
Unfurtuneally I deleted the nfs-mount in the PVE-Site and so it is not in the storage.cfg anymore.
So I have to disconnect a non existing nfsserver but hard connected nfs-share without reboot.

thx
Gerd

 

[YellowShed]

Hi,

I had a similar thing and found some help on this forum. I have manually modified the nfs mount in the storage.cfg file to add "soft,intr" to the options line which, as I understand from the documentation causes the nfs mounts to soft fault and interrupt the process calling it if the nfs mount goes away. I then re-started the nfs-common service. I haven't yet tried it to see if removing the nfs-mount causes any problems. It would be nice to have this option in the gui when mounting an nfs device.

YellowShed

 

[ggschmidt]

Hello YelloShed,

thanks for your post. I found this too and it will help me for the future. My main problem was how proxmox gets the shares, as I thought it should be via /etc/fstab. And how I can savely disconnect the missing nfs. Maybe via
"service portmap stop"? In Suse I could do "service nfs stop" as there is a extra ?? daemon for nfs-clients.

thx
Gerd

 

[YellowShed]

Hi Gerd.

Unfortunately (from another post about nfs mount missing the async option) I ran "cat /proc/mounts" and saw that the nfs mount still has "hard" in it and no "intr" option, so I think these have been ignored.

I'm sorry, I don't know how to safely unmount the missing nfs. I ended up hard-rebooting, but you could try the "service nfs-common stop"

I hope someone can confirm how to add options that don't get missed from the storage.cfg file or better yet, from the gui...

YellowShed

 

[Phlogi]

Please have a look here as well: https://forum.proxmox.com/threads/nfs-share-umount-f-problem-device-is-busy.26777/#post-136320