[Linux-cluster] Cluster NFS causes kernel bug

Ward, Timothy - SSD Timothy.Ward at itt.com
Tue Oct 23 18:57:15 UTC 2007


On Tue, 23 Oct 2007, Gordon wrote:

>> NFS 1.0.9-16 (BAD)
>> 
>> So I am avoiding Fedora and GFS2 until you've had more time to "bake"
it.
>> 
>> I am reading up on Wendy Chang's NFS linux-cluster postings to
>> understand why NFS fail over is not consistent.
> 
> What do you mean by "consistent"?
> 
>> - Has anyone successfully setup a failover NFS cluster?
> 
> You'll have to reveal more about your setup. Are you using a floating
IP and HA to fail over the NFS service, with GNDB for the mirrored block
device? A > SAN with a shared GFS file system with multiple nodes
exporting the NFS file system? Or something entirely different? 
> 
>> - Does nfsv4 work better?  It appears by default fedora6 uses nfsv3
(tcp) client and
>>   is flaky.
> 
> NFS over TCP has always been flaky, and from what I've seen NFSv4 is
not really any better. I'm still using NFSv3 over UDP for everything, as
it seems > > most reliable. 
> 
> Gordan

Gordon,

More detail:
- SAN FibreChannel array connected to a SAN switch with 2 of the 3 nodes
connected to it as well
- GFS1 fs mounted on the 2 nodes (So both SMB and NFS can access the
same files)
- A floating IP for the NFS service.  The failover domain only allows it
to go to one of the 2 SAN connected nodes
- A Samba service setup on the 2nd node but able to fail to the first
- NFS running on both nodes at all times, but only being cluster served
via the floating IP by one node

1) By "not consistent" I mean that when I failover from 1 node to
another node the NFS service fails just fine.  Its the client connection
that is painful.  Sometimes the client (FC6 in this case, building a
CentOS 5 as we speak) reconnects to the server "immediately", sometimes
within minutes, or an hour or never.

2) Thanks for the report on NFSv3/UDP.  From my reading that sounded
like something to avoid, but maybe I need to try it anyway.  How
reliable has it been?  Do the clients reconnect most times?

Thanks for the reply :)
Tim
*****************************************************************
This e-mail and any files transmitted with it may be proprietary 
and are intended solely for the use of the individual or entity to 
whom they are addressed. If you have received this e-mail in 
error please notify the sender. Please note that any views or
opinions presented in this e-mail are solely those of the author 
and do not necessarily represent those of ITT Corporation. The 
recipient should check this e-mail and any attachments for the 
presence of viruses. ITT accepts no liability for any damage 
caused by any virus transmitted by this e-mail.
*******************************************************************





More information about the Linux-cluster mailing list