Re: mount -a vs glusterfs

[Dima Veselov <kab00m%lich.phys.spbu.ru@localhost>]

On 13.09.2020 15:09, Michael van Elst wrote:

> > and one make "mount -a" several times - filesystem will
> > be mounted over and over
>
> > Filesystem                  Size       Used      Avail %Cap Mounted on
> > /dev/dk0                    993M       173M       770M  18% /
> > gfs01:/gva                  3,9T        45G       3,9T   1% /gva
> > gfs01:/gva                  3,9T        45G       3,9T   1% /gva
> > gfs01:/gva                  3,9T        45G       3,9T   1% /gva
>
> > unlike FFS filesystems which are not mounted again if
> > mounted before. Is it a bug and where it can reside?
>
> You can mount several filesystems on top of each other. Usually
> only the last mount is visible unless you mount with -o union
> where the next filesystems on the stack shines through.
>
> Mounting the same local filesystem like FFS twice fails because
> the underlying disk device only allows a single opener.
>
> But this doesn't happen here because the mount command, when started
> with -a option, checks the existing mount list if a mount point is
> already used by a filesystem of the same type. If you run mount
> with additional option -v, you will see which "already mounted"
> filesystems are skipped.
>
> Apparently this check fails for glusterfs and there is no
> other obstacle that prevents further mounts, so you get
> multiple mounts.

You're right, it says nothing about glusterfs "already mounted", just
tries to mount it over.

> The source suggests that glusterfs uses the filesystem type
> "puffs|perfuse|fuse.glusterfs" and the check above strips
> the "puffs|" prefix. Sounds like lots of magic, but not
> enough to make the check work.
>
> Please run 'mount' or 'mount -v' to see what filesystem type
> is actually recorded. "glusterfs" or "puffs!glusterfs" should
> be fine.

It says:
gfs01:/gva on /gva type puffs|perfuse|fuse.glusterfs

I'm confused - do you say the source check for phrase exactly
what mount show to me?

--
Sincerely yours,
Dima Veselov
Physics R&D Establishment of Saint-Petersburg University