On Sun, Aug 21 2016, Karsten Hohmeier wrote:
> Dear NFS list members,
>
> I recently contacted a few active NFS devs with an issue i ran into:
>
> We run a XenServer 6.5 (Kernel 3.10) with an older version of nfs-utils
> (1.0.9). NFS mountpoints on that machine acting as a client can not be
> unmounted (not even forcibly or lazy) after the connection to the NFS
> server broke for some reason.
>
> Most commands fail with "Stale NFS file handle".
>
> The Internet tells me this situation is unfixable and requires a reboot
> of the client to resolve. This is a pretty bad solution on a heavily
> loaded virtualization host.
>
> The response of Bruce Fields was: "Sounds like normal behavior, ..."
> (Sorry for not posting to the list in the first place, btw.)
>
> I'm still looking for some explanation why this happens, how it can be
> considered "normal behavior" and if it can be fixed/avoided/mitigated
> without rebooting.
This was fixed 3 years ago in Linux 3.12
Commit: 8033426e6bdb ("vfs: allow umount to handle mountpoints without revalidating them")
(and probably a few other related patches).
You need to be sure you have a version of util-linux where "umount -f"
and "umount -l" don't stat() the mount point.
NeilBrown
>
> Best regards
>
> Karsten Hohmeier
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at http://vger.kernel.org/majordomo-info.html
Attachment:
signature.asc
Description: PGP signature
