Scalix with DR Failover

[scalixnub]

I've done a little reading about clustering with scalix but haven't found any good info on running a master slave scalix solution.

ie. I have a master server at one site with its own db and I want an identical server at another site that is effectively my Disaster Recovery mail server with its own db. So the Mailstore and ldap are synced to the remote site over a 10mb link. So if my main site goes down I do a dns update and people connect to the DR server.

Has anyone done this with scalix? Anyone got some docco on it?

In my case a cluster is not an option as I would be relying on the same database being up.
Can you configure ldap replication to the slave and posgresql db replication?

Also has anyone tried a master-master config using this sort of configuration? ie. round-robin dns

[LeslieW]

To my knowledge nobody has done this.

You can set up a slave LDAP server at the remote location.
You don't need to replicate postgres; it's only used as a message header cache and will rebuild itself if necessary.

In your situation I would recommend installing scalix on the remote server, taking care that the uid and gid for the scalix user are the same on both servers. Then rsync /var/opt/scalix from your production server to your standby server each night.

I also wouldn't change DNS; I'd use the same IP on both servers. Otherwise every time you bring the standby server online you'll have to go through the steps outlined in the Wiki article about changing the IP of your server.

There are probably a lot of little surprises waiting for you, so be sure to test things thoroughly.

You could also consider setting up a cluster, and rsyncing the data store from the production location to the standby location. That way the failover software takes care of the IP issues.

A lot of people talk about setting up a remote DR server, in case of tornado, flood, fire, etc. but I've not seen anyone talking about an installed, configured, working solution and its strengths and weaknesses. This is a great idea for a thread and I hope to see some good activity on it.

[garthgza]

I am also interested in running a similar configuration and was thinking of doing the following:

The master Scalix server is set to Internal IP eg 10.0.1.202
The Slave DR server is set to internal IP eg 10.0.1.200

Both servers have the same hostname scalix.mycompany.com

I was thinking of using Rsync to sync the mail store (Approx 8Gb) from master to slave on a daily basis. Is this possible without stopping the scalix services first or is that not advisable?

What other things should I be aware of besides the importance of changing the IP address if both scalix servers are running Centos 5.3 and Scalix 11.4.3?

Your help is much appreciated.



My question is

[cacti4x4]

I have a setup running using LVM snapshots and rsync allowing me to get my DR server up and running inside 1 minute. The DR server is a complete duplicate of the production server.
Currently I sync e-mail once every hour... Check out http://www.scalix.com/forums/viewtopic.php?f=2&t=13676 for full details. My mail store is 250GB big.

It works so well I’m using this method to switch between the production and DR server (and back again) when ever I want to do hardware maintenance.

[cacti4x4]

I forgot to mention. If your brave it will be possible to write scripts to do an automatic failover once failure of the production server is detected.