SUMMARY: Sun Clusters

From: Gary Carr (
Date: Thu Jun 03 1999 - 14:35:07 CDT

Sorry to take so long, but I got a number of replies and I wanted to make
sure I had them all before sending summary.

Below is my original message:

> We are thinking of setting up a Sun Cluster to provide high availability
> for our NFS file server. The proposal is to use two Enterprise 2300 servers
> dual ported to an A1000 hardware RAID array using the Sun cluster software.
> We will probably use 100Base-T for the private communications path between
> the two servers.
> Does anyone have any experience with the Sun Cluster software/hardware? The
> Sun sales people assure us that in this setup if one of the servers dies
> the other will automatically take over all NFS file mounts and service them
> "seamlessly". Any experience with this situation?

In general the responses were positive. I won't try to include every
response, but here is a summary of what struck me as the most important points:

1) Sun Clusters work well. The base module, HA-NFS, which is what we are
primarily interested in, is the most reliable module.

2) They are complex to initially plan, setup, and install, and they are
complex to manage. I had several suggestions to have Sun do the initial
install, and to take the Cruster Management course that Sun offers.

3) Several respondents thought they were relatively expensive.

4) "Seamlessly..." may not be quite the appropriate term for how NFS file
mounts fail over to the backup server. It usually takes less than 2 minutes
to fail over, but in some cases it can take quite a bit longer. If the main
server fails in a particularly nasty way the backup server may have to fsck
all the disks. To the client workstations a fail over looks very similar to
a server reboot.

5) Since NFS is a stateless protocol there are usually no software problems
with fail over, but there can be problems with locks. I also had one reply
claiming they had some problems with automounter.

6) There are several alternatives to Sun Cluster software:
        Legato QHA+
        Veritas file system and HA (several very positive comments)
        Clarion fail over software package

Many thanks to the managers who were kind enough to respond to my original

This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:13:20 CDT