How to mount CIFS reliably on host, so bind mounts don't break?

General
1

We are using a Hetzner Storagebox and about every week the mount breaks on the Debian servers:

root@server1 /data # ll
ls: cannot access 'storagebox': Host is down
total 48
drwxr-xr-x  6 root root  4096 Dec 17  2022 .
drwxr-xr-x 19 root root  4096 Jan 31 23:16 ..
d?????????  ? ?    ?        ?            ? storagebox

root@server1 /data # mount.cifs -o vers=3,seal,user=u300000,pass=abcdef //u300000.your-storagebox.de/backup /data/storagebox

root@server1 /data # ll
total 48
drwxr-xr-x  6 root root  4096 Dec 17  2022 .
drwxr-xr-x 19 root root  4096 Jan 31 23:16 ..
drwxr-xr-x  2 root root     0 May  8 13:47 storagebox

I suspect that Hetzner interrupts the connection for updates of software or changes to hardware. I don’t understand why the mount is not automatically re-established by Debian - that’s what I would expect.

To add to the problem, even when the mount is re-created, any bind mount into Docker containers is not re-established, so all containers need to be restarted, kind of bad for high availability.

(Note: the Storagebox only allows 10 connections, so I can not connect every container directly to the Storagebox, but need to go through the host and use bind mounts.)

So as the title: How to mount CIFS reliably on host, so bind mounts don’t break?

2

Have you configured bind propagation for your binds?

Default behavior:
When a container is started, it will not pick up any changes to mounted folders within the bound folder. The same is true in case a mount is temporarily unavailable, and re-mounted later again.

The mount propagation types slave and rslave should propagate changes to submounts into the container, so that it picks up on the changes and reflects them.

See:

There is no guaranty that it won’t break, but at least the container can recover once the mount is working again.