Re: clients cannot reconnect to mounted nfs shares after server restart

["J. Bruce Fields" <bfields@fieldses.org>]

On Fri, Feb 04, 2011 at 01:31:05AM -0800, camden lindsay wrote:
> Thanks for the response:
> I have waited quite some time-- many minutes-- and still cannot reconnect.

The wait shouldn't be more than a minute or two after the nfs server is
back up and running.  And mount should pretty much never see a stale
filehandle error.

A network trace showing the stale errors after reboot, together with the
output of

	cat /proc/net/rpc/nfsd.*/content
	blkid

on the server, could be useful.

--b.

> 
> Server --> BigBox dylan # uname -a
> Linux BigBox 2.6.36-gentoo-r5 #5 SMP Tue Feb 1 06:57:03 PST 2011
> x86_64 Intel(R) Core(TM) i3 CPU 540 @ 3.07GHz GenuineIntel GNU/Linux
> 
> What i have found is that, with a newer client, i get slightly more
> information at the command line, pointing to stale nfs filehandle:
> Client1:
> Linux shobuntu 2.6.24-28-generic #1 SMP Wed Nov 24 09:30:14 UTC 2010
> i686 GNU/Linux
> mount.nfs: timeout set for Fri Feb  4 01:19:37 2011
> mount.nfs: text-based options: 'addr=192.168.11.10'
> mount.nfs: text-based options: 'addr=192.168.11.10'
> mount.nfs: text-based options: 'addr=192.168.11.10'
> mount.nfs: text-based options: 'addr=192.168.11.10'
> mount.nfs: text-based options: 'addr=192.168.11.10'
> 
> Client2:
> Linux core2buntu 2.6.32-28-generic #55-Ubuntu SMP Mon Jan 10 23:42:43
> UTC 2011 x86_64 GNU/Linux
> mount.nfs: timeout set for Fri Feb  4 01:18:17 2011
> mount.nfs: text-based options: 'addr=192.168.11.10'
> mount.nfs: mount(2): Stale NFS file handle
> mount.nfs: text-based options: 'addr=192.168.11.10'
> mount.nfs: mount(2): Stale NFS file handle
> mount.nfs: text-based options: 'addr=192.168.11.10'
> mount.nfs: mount(2): Stale NFS file handle
> 
> 
> Should this automatically converge at some point?
> If so, how long should i expect for convergence?
> 
> Thanks!
> Camden
> 
> On Tue, Feb 1, 2011 at 10:15 AM, J. Bruce Fields <bfields@fieldses.org> wrote:
> >
> > On Mon, Jan 31, 2011 at 07:24:30PM -0800, camden lindsay wrote:
> > > Having an issue i believe is an NFS bug, but would like someone to
> > > verify as well as... point me the correct direction to submit a bug as
> > > needed if it indeed is.
> > > Situation:
> > > NFS server running
> > > Client connected with mounted share
> > > NFS server is shut down -- (halt or /etc/init.d/nfs stop)
> > > NFS server is turned on (booted or /etc/init.d/nfs start)
> > >
> > > Outcome:
> > > NFS client that was connected does not reconnect to that share automatically
> > > NFS client that was connected cannot reconnect via umount then mount again
> > > No other nfs client can connect to that share, before or after origial
> >
> > How long have you waited?
> >
> > > NFS clients attempts to connect/reconnect
> > > NFS clients CAN connect to a different share.
> > > Details/logs:
> > > Server shows that client is authenticating:
> > > Feb  1 03:01:47 BigBox mountd[6557]: authenticated mount request from
> > > 192.168.11.49:674 for /mnt/store/public (/mnt/store/public)
> > > Feb  1 03:01:48 BigBox mountd[6557]: authenticated mount request from
> > > 192.168.11.49:763 for /mnt/store/public (/mnt/store/public)
> > > Feb  1 03:01:50 BigBox mountd[6557]: authenticated mount request from
> > > 192.168.11.49:932 for /mnt/store/public (/mnt/store/public)
> > > Feb  1 03:01:54 BigBox mountd[6557]: authenticated mount request from
> > > 192.168.11.49:942 for /mnt/store/public (/mnt/store/public)
> >
> > A network trace (tcpdump -s0 -wtmp.pcap, then reproduce the problem,
> > then ^C tcpdump, then send us tmp.pcap) might help.
> >
> > >
> > > Client shows retransmission:
> > > dylan@shobuntu:~$ sudo mount -v -t nfs bigbox:/mnt/store/public Public/
> > > mount.nfs: timeout set for Mon Jan 31 19:09:32 2011
> > > mount.nfs: text-based options: 'addr=192.168.11.10'
> > > mount.nfs: text-based options: 'addr=192.168.11.10'
> > > mount.nfs: text-based options: 'addr=192.168.11.10'
> > > mount.nfs: text-based options: 'addr=192.168.11.10'
> > > mount.nfs: text-based options: 'addr=192.168.11.10'
> > >
> > > Server Version:
> > > net-fs/nfs-utils
> > >       Latest version available: 1.1.4-r1
> > >       Latest version installed: 1.1.4-r1
> > >       Size of files: 574 kB
> > >       Homepage:      http://linux-nfs.org/
> > >       Description:   NFS client and server daemons
> > >       License:       GPL-2
> > > (server is on Gentoo, client on Ubuntu)
> >
> > Note most of the NFS code is actually in the kernel, so the kernel
> > version(s) involved would be useful too.
> >
> > --b.
> >
> > > Thanks!
> > > Camden
> > > --
> > > To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> > > the body of a message to majordomo@vger.kernel.org
> > > More majordomo info at  http://vger.kernel.org/majordomo-info.html