SUMMARY: NFS Client Failover

From: Charles Brian Hill (
Date: Thu Dec 21 2000 - 13:40:07 CST

My original question:

I'm curious about the NFS Client Failover that Sun introduced with
Solaris 2.6. I have several questions about it. I've searched various
Sun sites, the archives, and the Internet in general, and can't seem to
find much helpful information about NFS client failover. With that in
mind, if anyone has experience using the feature and can provide insight,
I'd be extremely grateful.

1) The mount_nfs(1m) man page specifies that failover should only be
   used for read-only mounts, using replicated filesystems. Does anyone
   know what happens if you mount a filesystem read-write, but specify
   multiple NFS servers, as is the case when you use the client failover
   feature? Specifically, I'm asking whether the client ignores the
   multiple servers, or will attempt to perform the failover even though
   the filesystem is mounted read-write.

2) Can the NFS client failover feature be used to handle failure of a
   NIC in one NFS server? In other words, if I have one server with
   multiple network interfaces, can I specify both interfaces of that
   server when I mount the filesystem?


To summarize, it seems that the OS is smart enough not to attempt
dynamic failover if the filesystem is mounted read-write. It will,
however, use the multiply defined hostnames at mount time, and if the
first listed server is unavailable, it will try the next in the list.

The answer to question 2) was, yes, that this can be done, obviously
with the restriction that the filesystem must be mounted read-only.

Thanks to the following people for their responses:

Casper Dik
Christophe Dupre
Clifford Thurber
Darren Dunham
David Foster

Charles Brian Hill

_______________________________________________ sunmanagers mailing list

This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:14:25 CDT