Hetzner Storagebox cifs error after IP change

[Enthylsa]

Hi, i am using a Hetzner storage box in my proxmox cluster. From time to time the connection is failing and I delete and add the storage box again to make to working.

Today I had the same problem and tried to find the root cause.

What I can see when calling mount is that the ipv4 address is used to mount the storage box:

//XXXX.your-storagebox.de/backup on /mnt/pve/storage-proxmox1 type cifs (rw,relatime,vers=3.1.1,cache=strict,username=XXXX,uid=0,noforceuid,gid=0,noforcegid,addr=65.21.90.67,file_mode=0755,dir_mode=0755,soft,nounix,serverino,mapposix,rsize=4194304,wsize=4194304,bsize=1048576,echo_interval=60,actimeo=1)

But if I dig the DNS for my storage box I receive a different IPv4 address. Seems that Hetzner change the IP adress for storage boxes from time to time.

Because I do not use the IP address but the FQDN for adding the storage to my cluster: What would be a proper solution? I a quite surprised that mounting the storage box via FQDN leads to a static ip mounting.

 

[fweber]

Hi, I suppose mount.cifs resolves the hostname an IP address once when it is called by PVE, and from then on, it operates on this IP address. I'm not aware of a way to configure it to resolve again when the connection fails.

One workaround I can think of: You could have a script periodically check whether the IP of the current mount is still up-to-date, and if it isn't, unmount the mount point (via umount -l /path/to/mountpoint). It would then take PVE a few seconds to detect that the storage is missing and run mount again, which should then resolve the hostname to the new IP.

 

[Enthylsa]

Hi I did unmount as you suggested and now it is using the ipv6 address. But I can recall that the IPv6 address was also used the last time. So there seems a certain point in time where the mount changes to ipv4. The server has not been restarted or so so I can't think of an event which cause the mount to switch from the IPv6 to IPv4 address...

 

[Enthylsa]

I created a shell script which I run as cronjob once a day to test and unmount. Here it is if anybody else has need for it:

#!/bin/bash

mount_point="/mnt/pve/{your_storage_name}"

get_timestamp() {
    date +"%Y-%m-%d %H:%M:%S"
}

df_output=$(df -h | grep $mount_point)

if [ -z "$df_output" ]; then
    echo "$(get_timestamp): Host is down. Remounting the volume..."

    umount -l $mount_point

    if [ $? -eq 0 ]; then
        echo "$(get_timestamp): Volume unmounted successfully."
        sleep 30

        df_output=$(df -h | grep $mount_point)
        if [ -n "$df_output" ]; then
            echo "$(get_timestamp): Volume has been remounted successfully."
        else
            echo "$(get_timestamp): Failed to remount the volume."
        fi
    else
        echo "$(get_timestamp): Failed to unmount the volume."
    fi
else
    echo "$(get_timestamp): Host is up. Volume is already mounted."
    fi