[ubuntu] Failover Clustering

[ICTNUT]

Right Chaps,

 

Need advice on providing faiover for our Ubuntu servers running our VLE.

 

A bit of background:

 

We have our nix setup as VMs on a rather hefty VMWare, Blade, SAN setup which means we would be looking to cluster virtual nix servers.

 

Anyone done it yet?

 

Any pitfalls?

[Geoff]

Here you go.

 

GettingStarted/Apache: Linux HA

 

The issue of database avalibility or sharing of the web root or the apache configuration is not considered.

[ICTNUT]

Yes at the moment we have the DBs and the web front end on the same server but I am looking at having 2 Apache VMs and 2 MySQL VMs with the MySQL VMs possibly replicating but I need to look into this abit more.

 

At the moment lets get the front ends running first

 

I maybe back

[Zimmer]

Good question ICTNUT, I wouldn't mind knowing how to do this myself.

 

I also have a number of Apache/MySQL web services that need to be placed in a failover cluster. For the time being I have just been doing nightly Robocopies and if a failure occurs I will change DNS to the backup server. Could do with something proper though to be honest!

[james_yale]

You might look at using LVS and heartbeat to provide a load balanced cluster, rather than having one server idle. It requires provisioning a load balancer (which should also have a failover partner) so is some extra work, but once you've got them up and running you can add extra services easily.

 

MySQL has built in replication that works really well (when it is working), the slave just follows the binary logs of the master. We're having trouble here however where the replication slave keeps stopping, though this is possibly only because the little virtualised replica can't keep up with the monstrously fast master server.

[Joedetic]

DRDB and Linux HA would be your best bet.