[Pacemaker] Problem to support NFS v3

Fri Sep 17 16:44:50 UTC 2010

Thanks Dejan.

I will try to look for answers from other components.

Liang Ma
Contractuel | Consultant | SED Systems Inc. 
Ground Systems Analyst
Agence spatiale canadienne | Canadian Space Agency
6767, Route de l'Aéroport, Longueuil (St-Hubert), QC, Canada, J3Y 8Y9
Tél/Tel : (450) 926-5099 | Téléc/Fax: (450) 926-5083
Courriel/E-mail : [liang.ma at space.gc.ca]
Site web/Web site : [www.space.gc.ca ] 

-----Original Message-----
From: Dejan Muhamedagic [mailto:dejanmm at fastmail.fm] 
Sent: September 17, 2010 5:27 AM
To: The Pacemaker cluster resource manager
Subject: Re: [Pacemaker] Problem to support NFS v3

Hi,
On Thu, Sep 16, 2010 at 09:04:29AM -0400, Liang.Ma at asc-csa.gc.ca wrote:
> Somehow this may not get through. Let me try it again.
> 
> Hi There,
> 
> I have setup a pair of primary/secondary servers for services
> such as web, ftp, samba and nfs with pacemaker 1.1.1,
> drdb-pacemaker 8.3.7, corosync 1.2.7 on Fedora 13. Things work
> well except for an old Solaris 2.7 client. The Solaris system
> can't mount to the nfs server because it can support NFS v3 not
> v4. From tcpdump this is what goes wrong with a pure NFS v3
> client,
> 
> 19:55:02.886062 192.168.249.150.39331 > 10.1.1.200.sunrpc:  udp 84 (DF)
> 19:55:02.886625 10.1.1.199.sunrpc > 192.168.249.150.39331:  udp 28 (DF) 
> 20:01:12.741180 192.168.249.150.52440 > 10.1.1.200.sunrpc:  udp 84 (DF)
> 20:01:12.741667 10.1.1.199.sunrpc > 192.168.249.150.52440:  udp 28 (DF)

> 
> Where 10.1.1.200 is the virtual IP address and 10.1.1.199 is the IP address of the primary server. If I try to mount to the non-virtual address 10.1.1.199, it works fine. But obviously this is what we want with a virtual server.
> 
> Originally I used resource agent lsb:nfs. Then changed to ocf:heartbeat:nfsserver with specific parameter pointing to the virtual address 10.1.1.200. Plus I added rpcbind which takes care sunrpc request as a virtual service too (see the related configuration below please). It has the same thing as before: The virtual server responds with its physical address. 
> primitive nfs ocf:heartbeat:nfsserver \
> 	params nfs_ip="10.1.1.200" nfs_shared_infodir="/var/drbdata0/exports" \
> 	params nfs_init_script="/etc/init.d/nfs" nfs_notify_cmd="/sbin/rpc.statd" \
> 	op monitor interval="5s" timeout="20s" depth="0" \
> 	meta target-role="Started"
> primitive rpcbind lsb:rpcbind \
> 	op monitor interval="50s" \
> 	meta target-role="Started"
> 
> The virtual NFS service works fine with nfs clients that
> support NFS v4. Any ideas please?

I can vaguely recall problems like this in the past, but missing
all the details :-/ At any rate, the problem is related either to
the kernel or NFS. I think you should better ask in the Fedora or
the NFS related forum.

Thanks,

Dejan

> Thanks in advance.
> 
> 
> Liang Ma
> Contractuel | Consultant | SED Systems Inc. 
> Ground Systems Analyst
> Agence spatiale canadienne | Canadian Space Agency
> 6767, Route de l'Aéroport, Longueuil (St-Hubert), QC, Canada, J3Y 8Y9
> Tél/Tel : (450) 926-5099 | Téléc/Fax: (450) 926-5083
> Courriel/E-mail : [liang.ma at space.gc.ca]
> Site web/Web site : [www.space.gc.ca ] 
> 
> 
> 
> 
> 
> _______________________________________________
> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
> 
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
> Bugs: http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker

_______________________________________________
Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
http://oss.clusterlabs.org/mailman/listinfo/pacemaker

Project Home: http://www.clusterlabs.org
Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
Bugs: http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker