[Samba] CTDB and NFS4

[Volker Lendecke]

On Tue, Mar 11, 2014 at 09:09:15PM +1300, Stephen Ward wrote:
> Hi Volker, thank you for your comments.
> 
> Using CTDB with an NFS server if I mount with the version 3 option then
> stop ctdb on the server the client fails over to the new server fine.  Same
> client, same servers, same process but mounting with the version 4 option
> and when CTDB fails over the client locks.  After ten minutes or so I
> killed the process on the client, it was still not responsive.
> 
> As I understand it this works because the NFS daemons share records and
> 'tickle' connections on failover (see
> https://wiki.samba.org/index.php/CTDB_Setup#Setting_up_CTDB_for_clustered_NFS
> ).
> 
>  NFS4 works fine except for failover and NFS3 works fine including failover
> so I assume that it is specifically an issue with NFS4 with CTDB.
> 
> My understanding of specifically how this works is limited so if anyone has
> any ideas about this at all I'd be really keen to hear!

Ok. Understood. The problem here is that we (Samba/ctdb
team) have no real connection to the NFS4 people except for
some random chats with the Ganesha team. But I guess you're
not using Ganesha anyway. NFS4 is probably a lot more
difficult to correctly fail over in a cluster case, because
it is stateful and does have a notion of open files. So I'm
not sure how the clients in your setup are supposed to work,
and how Samba/ctdb could improve things.

Do you have an idea how this is supposed to work in a
Pacemaker failover scenario, no ctdb and Samba involved,
just NFS4?

Volker

-- 
SerNet GmbH, Bahnhofsallee 1b, 37081 Göttingen
phone: +49-551-370000-0, fax: +49-551-370000-9
AG Göttingen, HRB 2816, GF: Dr. Johannes Loxen
http://www.sernet.de, mailto:kontakt at sernet.de