:: Re: [DNG] NFS Stale file handle for…
Top Pagina
Delete this message
Reply to this message
Auteur: Jackman
Datum:  
Aan: g4sra
CC: DNG
Onderwerp: Re: [DNG] NFS Stale file handle for regular user, not root
Mounting and remounting is not the problem, unfortunately. That'd be easier
to track down. I can't unmount with the regular user, but root can do it
just fine.

BTW, the 'jackman' user has the same uid on both server and client.

On Wed, Feb 24, 2021, 3:39 PM g4sra via Dng <dng@???> wrote:

> ‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐
> On Wednesday, February 24, 2021 10:03 PM, Jackman via Dng <
> dng@???> wrote:
>
> I don't even know how to Google for this.
>
> Hosts involved:
> storage0: NFS/KRB5 server
> dorito: NFS client
>
> As root, I can mount, list files, view files, and so on. When I try to
> list files on the mount, I get a stale file handle error.
>
> ➜  ~ mount /mnt/backup && ls -hal /mnt/backup
> ls: cannot open directory '/mnt/backup': Stale file handle
> ➜  ~ ls /mnt -hal
> ls: cannot access '/mnt/backup': Permission denied
> total 24K
> drwxr-xr-x  7 root root 4.0K Feb 24 12:10 .
> drwxr-xr-x 22 root root 4.0K Feb 23 15:56 ..
> d?????????  ? ?    ?       ?            ? backup

>
> Note the difference with root:
>
> dorito :: ~ » mount /mnt/backup && ls -hal /mnt/backup
> total 57K
> drwxr-xr-x 12 root    root      12 Feb 23 03:36 .
> drwxr-xr-x  7 root    root    4.0K Feb 24 12:10 ..
> drwxr-xr-x  3 jackman jackman    3 Feb 11 04:21 backup_dorito_20210211
> drwxr-xr-x  4 jackman jackman    4 Feb 23 06:16 dorito_20210223-0336
> dorito :: ~ » ls -hal /mnt
> total 26K
> drwxr-xr-x  7 root root 4.0K Feb 24 12:10 .
> drwxr-xr-x 22 root root 4.0K Feb 23 15:56 ..
> drwxr-xr-x 12 root root   12 Feb 23 03:36 backup

>
> Cute, eh?
>
> This is not a problem on the storage server itself:
>
> storage0 :: /srv » mount /mnt/backup && ls -hal /mnt/backup
> total 57K
> drwxr-xr-x 12 root    root      12 Feb 23 03:36 .
> drwxr-xr-x  8 root    root    4.0K Feb 24 14:16 ..
> drwxr-xr-x  3 jackman jackman    3 Feb 11 04:21 backup_dorito_20210211
> drwxr-xr-x  4 jackman jackman    4 Feb 23 06:16 dorito_20210223-0336

>
> Here are some relevant things:
>
> ➜ ~ mount | grep nfs
> storage0:/srv/backup on /mnt/backup type nfs4
> (rw,nosuid,nodev,noexec,relatime,vers=4.2,rsize=1048576,wsize=1048576,namlen=255,hard,proto=tcp,timeo=600,retrans=2,sec=krb5,clientaddr=10.1.0.110,local_lock=none,addr=10.1.0.100,user=jackman)
>
> ➜  ~ cat /etc/fstab | grep backup
> storage0:/srv/backup    /mnt/backup     nfs4
>  noauto,rw,noexec,user,sec=krb5  0       0

>
> storage0 :: /srv » cat /etc/exports | grep backup
> /srv/backup     10.0.0.0/8(rw,no_subtree_check,sec=krb5)

>
> storage0 :: /srv » cat /etc/default/nfs-kernel-server
> RPCNFSDCOUNT=32
> RPCNFSDPRIORITY=0
> RPCMOUNTDOPTS="--manage-gids"
> NEED_SVCGSSD="yes"
> RPCSVCGSSDOPTS=""
> storage0 :: /srv » cat /etc/default/nfs-common
> NEED_STATD=
> STATDOPTS=
> NEED_IDMAPD="yes"
> NEED_GSSD="yes"
>
> Networking is static:
>
> storage0 :: /var/log/kerberos » cat /etc/hosts
> 127.0.0.1       localhost
> ::1     localhost ip6-localhost ip6-loopback
> ff02::1 ip6-allnodes
> ff02::2 ip6-allrouters
> 10.1.0.100      storage0.jackman.local storage0
> 10.1.0.110      dorito.jackman.local dorito

>
> The mount appears to execute cleanly:
>
> ➜ ~ mount -v /mnt/backup
> mount.nfs4: timeout set for Wed Feb 24 14:27:58 2021
> mount.nfs4: trying text-based options
> 'sec=krb5,vers=4.2,addr=10.1.0.100,clientaddr=10.1.0.110'
>
> I don't see anything in the system logs on either machine that look at all
> relevant.
>
> Nothing odd (IMHO) is happening in the KRB5 logs, just successful grants.
>
>
> I thought this problem was exclusive to dorito (the client), so I nuked
> and re-installed Devuan. I have since installed Xubuntu on my desktop and
> it has the same issue now, too, but I'm politely setting that machine aside
> as it's not a Devuan machine.
>
> BTW, I've tried with 'noac', with no apparent change in behavior.
>
> I'm happy to RTFM if I knew what I was looking for.
>
> As always, any helpful souls are welcome to solicit on behalf of their
> beer fund.
>
> Andrew Jackman
> kd7nyq@???
>
>
> The unmount/remount technique won't work if some process is locked to it.
> Use 'lsof' as root to check the share on both the server and the client.
> If you find something, kill it and try the unmount/remount again.
>
> NB: this advice is worth exactly what you paid for it :)
>
>
>
>
> _______________________________________________
> Dng mailing list
> Dng@???
> https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng
>