Failed mount leaves corrupt mountpoint directory

[gareththered]

If I attempt to mount a remote server via SSHFS and it fails (in my case, it was a typo in the username, so I pressed Ctl-C to cancel), the mountpoint directory is left in an unusable state. I'm using the command without any option - simply sshfs user@server:dir/to/share /mount/point. The mount command shows the mount listed, but a ls shows:

root@web1:/srv# ls -l
ls: cannot access 'encrypted': Transport endpoint is not connected
total 4
drwxr-xr-x 2 www-data www-data 4096 Apr  3 16:03 docs
d????????? ? ?        ?           ?            ? encrypted
root@web1:/srv#

I can fix it with umount /srv/encrypted and the original directory returns.

This is on Debian 9 with sshfs version 2.8-1 and openssh version 7.4p1-10+deb9u3

[Nikratio]

Thanks for the report! I think this is not a bug but a feature (at least sort of). Imagine that you have a directory sshfs mounted and periodically sync it to a local directory (e.g. using rsync -a --delete). If sshfs terminates due to some bug, you wouldn't want to see an empty mountpoint, because that would mean that the sync process also deletes all files in the mirror.

I agree that in your case this behavior is annoying, but it is difficult to distinguish this programmatically.