[CentOS] Fault tolerance with webservers

Wed May 28 00:26:14 UTC 2008
Ross Cavanagh <ross-cavanagh at bm-sms.co.jp>

sbeam wrote:
> Interested in this discussion too, for reasons recently discussed...
>
> On Tuesday 27 May 2008 09:07, Fajar Priyanto wrote:
>   
>> For a starter, there is a very simple tool for this.
>> It's http://www.inlab.de/balance.html
>>     
>
>   
>> 2. RedHat Cluster Suite dan Piranha (http://www.redhat.com)
>> 3. Linux Virtual Server (http://www.linuxvirtualserver.org)
>>     
>
> Do all of these (or IPVS or Cluster Suite/GFS) take care of real-time sharing 
> of storage (sessions, database, files, logs) between all nodes? For a LAMP or 
> JEE or any other HTTP stack serving anything but readonly static files, this 
> is usually a requirement. 
>
> GFS is for sharing filesystem I know and there are howtos. So would you put 
> Balance or LVS on top of GFS, or...
>
> Would HA/DRBD be on the short list? http://www.drbd.org/ in our case we have a 
> two-node cluster anyway so this seems like the most straightforward option. 
> Or would something else be superior, more up-to-date?
>
> Sam
I think for a two nodes with real time storage sharing heartbeat / drbd 
would offer what you want. There's a pretty simple how to on the centos 
wiki to get you up and running.

I've not really used the RedHat cluster suite, but I 'think' it can 
offer a similar solution.

http://wiki.centos.org/HowTos/Ha-Drbd

-Ross-