[Linux-cluster] NFS failover problem

[Maciej Bogucki]

Paul n McDowell napisał(a):
> 
> Hi Maciej,
> 
> I'm not sure about these recommendations.  I agree with Peter Kruse's
> comments about creating a symbolic link for /var/lib/nfs [1 point 4e].
>  I have multiple NFS services in my cluster each serving a different
> file system so what would I be soft linking /var/lib/nfs to?
> 
It doesn't matter how many NFS services You have. The state information
for the NFS server is in /var/lib/nfs, and You have to put it on shared
storage.

> And for the NFS statd recommendation [1 point 4f], again I have several
> NFS services running on my cluster, each with their own IP alias, so
> what should I be setting STATD_HOSTNAME to in /etc/sysconfig/network ?

Takene from "man rpc.statd"

---cut---
       -n, --name name
              specify a name for rpc.statd to use as the local hostname.
By default, rpc.statd will call gethostname(2) to get the local
hostname. Specifying  a  local  hostname
              may be useful for machines with more than one interfaces.
----cut---
	
> 
> I never had to make either of these changes in my previous NFS server
> running on a RHEL 3 cluster which worked perfectly.

First please to make sure that it doesn't help You, then seek different
problem. /var/lib/nfs not on stared storage is common problem on Fedora,
SUSE and Red Hat distribution.

Best Regards
Maciej Bogucki