GlusterFS with Heartbeat

[dfriestedt]

I'm researching various high availability options for zimbra. I know Zimbra supports the Red Hat option, but I was looking at alternatives.

I see here - some people have gotten DRBD and Heartbeat working. Can anyone report Zimbra NE in production on DRBD / Heartbeat.

Has anyone tried Zimbra on GlusterFS with Heartbeat? I've been poking around on the glusterfs web site, and their cluster software looks interesting. I'm going to get a test env't up and see how it performs.

[phoenix]

Please update your forum profile with the output of the following command:

zmcontrol -v

Originally Posted by dfriestedt

I see here - some people have gotten DRBD and Heartbeat working. Can anyone report Zimbra NE in production on DRBD / Heartbeat.

The only supported high availability for NE use is RHEL Cluster Services, no other HA is supported and you will not receive support for any other configuration.

[blason]

Understood if not NE then I believe building up a cluster using glusterfsand heartbeat can be a good option. Did you succeed in that? Do you have white papers on it or any other written material?

[vavai]

Originally Posted by blason

Understood if not NE then I believe building up a cluster using glusterfsand heartbeat can be a good option. Did you succeed in that? Do you have white papers on it or any other written material?

i've never trying active-active cluster with Heartbeat and GlusterFS but I do have active-passive cluster with DRBD+Heartbeat with ext3 FS.

Zimbra - Search

[blason]

OK - So vavai just curious to know the cluster that you built was that a single server or multi server cluster? And do you have a goot write up on that? I have around 4000 users gonna use Zimbra OSS and looking for good clustering option.

[vavai]

Hi,

Originally Posted by blason

OK - So vavai just curious to know the cluster that you built was that a single server or multi server cluster? And do you have a good write up on that? I have around 4000 users gonna use Zimbra OSS and looking for good clustering option.

I'm using single server for DRBD+Heartbeat. I have a docs for implementing Zimbra on DRBD+Heartbeat but the current documentation is written in Indonesian.

I'll be try to translating my docs and post link on this thread.

[blason]

So vavai, is that stuff tested in production environment? How many mailboxes it has? Just curious to know if the this setup can sustain in production.

[vavai]

Originally Posted by blason

So vavai, is that stuff tested in production environment? How many mailboxes it has? Just curious to know if the this setup can sustain in production.

Yes, it's on production server. Mailboxes are vary, between 100 to 5000 mailboxes (I have some clients with different number of mailboxes) with about 1K-100K emails/day

[blason]

Hey Vavai,

I just implemented your stuff and wow DRBD seems to be working. Now next challenge in front of me is to configure Heartbeat which I partially completed. Since this is CentOS 5.5 I created ha.cf, haresource and authkeys under /etc/ha.d. However I bit confused about haresource since I read that it can vary per linux flavor, so would you please let me know if below haresources file can suffice my need?

Here is my ha.cf file
**************

[root@mail1.kus.net ~]# more /etc/ha.d/ha.cf
keepalive 2
deadtime 30
warntime 10
initdead 120
bcast eth0
node mail1.kus.net
node mail2.kus.net
auto_failback on
crm yes

***********
authkeys
auth 1
1 sha1 MySecret

***********
haresources

[root@mail1.kus.net ~]# more /etc/ha.d/haresources
mail1.kus.net IPaddr::192.168.1.128/24/eth0 drbddisk::r0 Filesystem::/dev/drbd0::/opt::ext3 zimbra

[vavai]

Originally Posted by blason

Hey Vavai,
Here is my ha.cf file
**************

[root@mail1.kus.net ~]# more /etc/ha.d/ha.cf
keepalive 2
deadtime 30
warntime 10
initdead 120
bcast eth0
node mail1.kus.net
node mail2.kus.net
auto_failback on
crm yes

***********
authkeys
auth 1
1 sha1 MySecret

***********
haresources

[root@mail1.kus.net ~]# more /etc/ha.d/haresources
mail1.kus.net IPaddr::192.168.1.128/24/eth0 drbddisk::r0 Filesystem::/dev/drbd0::/opt::ext3 zimbra

Your configuration seems to be enough, but I worries your Zimbra system will not be transferred to secondary nodes because of Zimbra hostname unless you have proper initial configuration.

Below is some clue based on my experience :

mail1.kus.net : IP 192.168.1.126
mail2.kus.net : IP 192.168.1.127
mail.kus.net : IP 192.168.1.128

1. Installing Zimbra on Primary node with hostname mail.kus.net with IP 192.168.1.128
2. Shutting down Zimbra services and then revert setting on primary node (mail1.kus.net : IP 192.168.1.126)

3. Installing Zimbra on secondary node with hostname mail.kus.net with IP 192.168.1.128
4. Shutting down Zimbra services and then revert setting on secondary node (mail2.kus.net : IP 192.168.1.127)

5. Setting up DRBD as your setting

6. Setting up Heartbeat as your setting

Summary : make sure that you have installing Zimbra with Heartbeat IP (virtual IP given by Heartbeat) and proper hostname (mail, instead of mail1 or mail2).

Hope this helps you for simplify your setting.