how to kill old NFS connections

[Thomas k.]

Hi!
I have recently removed an old NFS server from the shared storage. Umounted it from /mnt/pve from all nodes with

 umount -l /mnt/pve/nfs

I found out that on one node there are still connections to the old NFS Server (192.168.1.43) . I saw that with tcpdump:

root@pve2:/mnt/pve# tcpdump -netti vmbr0 host 192.168.1.43
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on vmbr0, link-type EN10MB (Ethernet), capture size 262144 bytes
1667482187.144316 e4:11:5b:b2:8a:84 > 34:17:eb:bb:89:18, ethertype IPv4 (0x0800), length 174: 192.168.1.81.758 > 192.168.1.43.2049: Flags [P.], seq 962730525:962730633, ack 2487151477, win 11027, options [nop,nop,TS val 4124555961 ecr 2611264902], length 108: NFS request xid 176762681 104 getattr fh Unknown/0100070101000C0A00000000865F2978FD3B43F59756C49D15638C9AA305300C

What can I do to stop the PVE node from doing that? I really would like to avoid a reboot of the node.

Thanks!
Thomas

 

[Moayad]

Hello,

in order to umount the storage in Proxmox VE you have to do this from PVE UI or CLI tools.

Regarding the issue you have, I would check if the mount service is still exists by issuing `systemctl status mnt-pve-<MOUNT NAME>.mount replace the "MOUNT NAME" with the name of the umounted NFS storage.

If the service exists, you have to disable it and delete, or just remount the storage again and remove it from the PVE web UI.

 

[Thomas k.]

Hello Moayad,

the mount is already deleted and is not showing up anywhere anymore. The strange thing is that only this one Node is still communicating with the old NFS server, even thoug it is not mounted anymore. I think the network connection/socket does still exist. I s there a way to tear down this connection? Some daemon restart?

 

[Moayad]

Hello,

Does the `lsof -N` command show that the old NFS, still lists the old NFS storage name?

Does anything show interesting in the journalctl/syslog on that node?