Thoughts and Suggestions for fail over options?

[jwatson]

I have 4 Zimbra Network Edition servers total. One is production, one is "fail over" for production in the same data center. And 2 are in another data center across the country for " fail over" for the first ones.

No shared SAN disk, previous admin is rsyncing from production, to another, to another, to another.

There is a TSM backup of the servers nightly. They want zero down time as well. again.

What if the zmbackup was pointed to an NFS on one of the other servers?

The email administration hat has been thrown in my lap.

HELP!!!!

[Krishopper]

Well, the "supported" method would be to use a SAN, so you can quickly failover and see the same data.

If that's not an option for you, you can use DRBD for realtime replicaton betwee the nodes.

I'm not sure what/if anything, you are using for failover between nodes, or if you handle that manually.

If you are doing it manually, you will just need to switch the primary disk over to the failover node with DRBD and bring things up. You can also use the open source Linux HA tools, or even RedHat cluster to take care of the steps for you should the main server fail.

Regarding backups, backing up to an NFS location is a great option, but doing a restore will take some time if you are trying to bring the failover node up -- so I would still look at a SAN or disk replication options to bring the mailbox server back up asap.