Fix `beegfs-client stop` stuck when management server is down

[jteng2127]

Is your feature request related to a problem? Please describe.

When the management server is down, running umount ${mnt} hang indefinitely, causing the systemctl restart beegfs-client incorrectly reported success.

Here is the process to reproduce it:

1. systemctl start beegfs-mgmtd start remote management server
2. systemctl start beegfs-client (which calls /opt/beegfs/sbin/beegfs-client start)
3. systemctl stop beegfs-mgmtd stop remote management server
4. systemctl restart beegfs-client (which calls /opt/beegfs/sbin/beegfs-client stop and /opt/beegfs/sbin/beegfs-client start)
   /opt/beegfs/sbin/beegfs-client stop is killed due to timeout, getting stuck at this line:
   

   beegfs/client_module/scripts/etc/beegfs/lib/init-multi-mode.beegfs-client

   Line 260 in a0357ce

   res=`umount ${mnt} 2>&1`

   
   Here is the systemd log when restarting the client:

   Aug 21 10:30:11 pc systemd[1]: Stopping Start BeeGFS Client...
   Aug 21 10:30:11 pc beegfs-client[58947]: Shutting down BeeGFS Client:
   Aug 21 10:30:11 pc beegfs-client[58947]: - Unmounting directories from /etc/beegfs/beegfs-mounts.conf
   Aug 21 10:31:41 pc systemd[1]: beegfs-client.service: Stopping timed out. Terminating.
   Aug 21 10:31:41 pc systemd[1]: beegfs-client.service: Control process exited, code=killed, status=15/TERM
   Aug 21 10:31:41 pc systemd[1]: beegfs-client.service: Failed with result 'timeout'.
   Aug 21 10:31:41 pc systemd[1]: Stopped Start BeeGFS Client.
   Aug 21 10:31:41 pc systemd[1]: Starting Start BeeGFS Client...
   Aug 21 10:31:41 pc beegfs-client[58986]: Starting BeeGFS Client:
   Aug 21 10:31:41 pc beegfs-client[58986]: - Loading BeeGFS modules
   Aug 21 10:31:41 pc beegfs-client[58986]: - Mounting directories from /etc/beegfs/beegfs-mounts.conf
   Aug 21 10:31:41 pc systemd[1]: Finished Start BeeGFS Client.
   

5. After that, systemctl status beegfs-client reports SUCCESS, because stop doesn't unmount successfully, and start skip the mount if it already existed
   

   beegfs/client_module/scripts/etc/beegfs/lib/init-multi-mode.beegfs-client

   Lines 190 to 194 in a0357ce

   	mount -t beegfs | grep "${mnt} " >/dev/null 2>&1
   	if [ $? -eq 0 ]; then
   	# already mounted
   	continue
   	fi

Describe the solution you'd like

Changing this line to use umount -l (lazy unmount) should work, but I'm not sure if it will break something

beegfs/client_module/scripts/etc/beegfs/lib/init-multi-mode.beegfs-client

Line 260 in a0357ce

res=`umount ${mnt} 2>&1`

Describe alternatives you've considered

• umount -f can also do force unmount, but it might crash other process.
• use longer TimeoutStopSec to client service.