On Jan 15, 2008 3:55 PM, Mathieu Bruneau <[EMAIL PROTECTED]> wrote:
> Hi,
>  Question on the side:
>
> What is your preferred solution to have such a shared Partition (specifically 
> for Kannel here)?

drbd + heartbeat

drbd  (http://www.drbd.org/) will create a raid 1 over network. each
server will have the same partitions, from which one will be
synchronized with drbd, and will be mounted only on the active node.

heartbeat (http://www.linux-ha.org/) will perform the failover functions

Here are some links on setting drbd wiith heartbeat:
http://www.slackworks.com/~dkrovich/DRBD/  - drbd howto
http://gentoo-wiki.com/HOWTO_Heartbeat_and_DRBD
http://www.rhic.bnl.gov/hepix/talks/041020am/miers.pdf
http://wiki.openvz.org/HA_cluster_with_DRBD_and_Heartbeat
http://www.geocities.com/arhuaco/doc/uml/uml-drbd-nfs-heartbeat.html
http://thnetos.wordpress.com/2007/06/18/drbd-and-heartbeat-for-high-availability-on-linux/
http://www.linuxjournal.com/article/9074

If the work of the server depends on many software components you may
want to use some
software to monitor the state of those process, such as monit or mon.
If you choose this I suggest you to run drbd as a kernel module and to
configure heartbeat to cooperate with
monit (http://www.tildeslash.com/monit/)
or mon (http://mon.wiki.kernel.org/index.php/Main_Page).

Here is a link for setting heartbeat and monit to work together:
http://linux.die.net/man/1/monit

Here is a link for setting heartbeat and mon to work together:
http://www.geocities.com/latompa/ha/apache_heartbeat.html
http://www.linuxvirtualserver.org/docs/ha/heartbeat_mon.html

BR, Jovan

Reply via email to