You got it. Both servers serve the same site. If one dies the other
takes over for it. To answer your question about database data, use
the same database server from both nodes. That database server could
be yet another failover cluster.

The most popular way to do failover database (mysql, postgres) with
heartbeat is by using DRBD (distributed replicated block device) which
will sync the underlying volume that stores the mysql data between the
nodes. When the primary dies, the secondary can mount the data, start
the database server and you are back in business. I imagine the same
strategy could be used with sybase on linux.

For static content, look at rsync or unison. You could optionally use
drbd to replicate your web content between boxes. Or use NFS to share
the web content from another box altogether. And yes, you guessed it,
that NFS server could be yet another cluster :-).

When you plan your HA setup, you want to eliminate single points of
failure. If you have two web servers in a cluster using a database
server and the database server dies, so much for your HA web servers.
What about your network switch? If it dies, again, your clustered web
servers are no help, you are still down.

Point is, clustering a set of resources is just one piece of the HA
puzzle, you have to consider ALL the resources that are needed to
serve up your application and find ways to make them fault tolerant.

On Wed, Mar 11, 2009 at 3:27 PM, Dimitri Yioulos <[email protected]> wrote:
> Ben,
>
> I did as you said, turning off apache on both nodes 1 and 2.  When I put node
> 1 on standby, HA turned on httpd on node 2.  When I brought it out of
> standby, HA turned on httpd on node 1, and turned it off on node 2.  So,
> success!  Thank you very much!
>
> OK, I think the lightbulb just got switched on.  The Web sites on nodes 1 and
> 2 should be exactly alike.  If node 1 goes down, then node 2 kicks in,
> delivering the exact same content.  Right?  Doh!  (I'm going to ask for my
> college tuition money back :-) ).
>
> If I might press on, how would that work with a database (take any; we use
> sybase running on Linux here).  If the database on the "back up" server/node
> 2 is off, to be powered on if the database on the node 1 is unavailable, how
> is data replicated between the two?  I hope it's OK to ask this.
>
> Dimitri
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

Reply via email to