Fedora 20 nfs

[T.C. Hollingsworth]

On Mon, Oct 14, 2013 at 2:41 PM, Eric Blake <eblake at redhat.com> wrote:
> Dunno.  Whatever I got by installing F17 with anaconda then
> incrementally upgrading through F18, F19, and now F20, and where I set
> up my /etc/fstab by copying the same configuration that worked for me
> since F12 pre-systemd days.  So it might not be ideal, but it's one of
> those "it works well enough for me that I'm not going to waste time
> tweaking it unless it breaks first" situations.  Except that it's
> noticeably broken enough at shutdown that I bothered to ask on the list :)
>
> The corresponding /etc/fstab entry:
>
> nas:/backup      /mnt/backup          nfs     bg,user,_netdev 0 0

Okay, so the "bg" option is a little different from what most people
refer to as automounting, in that it just repeatedly attempts to mount
the share until it succeeds, whereas true automounting waits until you
attempt to access the mount to even try to mount it.  Of course, this
distinction matters very little to _you_, but it might indicate what
systemd is getting wrong here.

I'm curious as to whether systemd even tracks the mount properly in
this case.  Does `systemctl status mnt-backup.mount` indicate success
or failure?

If it indicates success, systemd definitely should be tearing down the
mount on shutdown.  (systemd by design is supposed to reverse
Before/After deps for stop operations.)  Definitely file a bug in this
instance.

If it indicates failure, systemd isn't getting informed that this
mount actually succeeds.  You could file a bug against systemd
regarding this, but their answer might just be "use real automounting
if you want this to work properly." To do that, switch your "bg" mount
option for "x-systemd.automount" and see if it gets unmounted properly
on shutdown afterwards.

Their answer could just as easily be "yeah, we need to fix this", so
please do file the bug anyway, if only for the benefit of others who
might run into this.

-T.C.