[SOLVED] Zimbra Cluster with drbd won't start

[tibby]

I'm building an Active/Passive zimbra cluster with DRBD.

I have installed Zimbra on two nodes.
(installed one node, then switched drbd to the other, deleted /opt/zimbra. It was mounted from /dev/drbd1, and did another clean install on the 2nd node.)

the /etc/hosts file are the same on both nodes:

cat /etc/hosts
127.0.0.1       localhost
XXX.XXX.XXX.XX4 mydomain.net
XXX.XXX.XXX.XX3 mydomain.net Node01
XXX.XXX.XXX.XX2 mydomain.net Node02

I am using heartbeat to switch resources, that's why i have XXX.XXX.XXX.XX4 mydomain.net configured as a virtual ip.

MX records set up like this:

mydomain.net IN MX  0 mydomain.net
mydomain.net IN MX 10 Node01.mydomain.net
mydomain.net IN MX 20 Node02.mydomain.net

The problem when I switch nodes, (reboot Node01), DRBD mounts /opt but when I want to start zimbra from /etc/init.d/zimbra start, it wont start on the second node. It's working fine on node 1, but the 2nd does not.
I know I should put it as a start up into the /etc/heartbeat/haresources but at the moment I would like to see the system working manually. Once I can start it on Node02 I'll put into heartbeat to start it, but until that I would like to see it working.

the error I'm getting:

/etc/init.d/zimbra start
Host localhost
Unable to determine enabled services from ldap.

Why does this wants to go for localhost?? It has the virtual IP, and Zimbra was configured to use that.

Thanks for the help! I've been strugling with this for two days now.

[phoenix]

Why does this wants to go for localhost?? It has the virtual IP, and Zimbra was configured to use that.

The usual reason for the LDAP error is that you don't have a valid A or MX record pointing to the server. You could also check out these posts: site:zimbra.com +drbd +"/etc/hosts" - Yahoo! Search Results

[tibby]

The usual reason for the LDAP error is that you don't have a valid A or MX record pointing to the server. You could also check out these posts: site:zimbra.com +drbd +"/etc/hosts" - Yahoo! Search Results

But if this would be the case, Zimbra would not work on any of the servers.
Node01 is working fine. it's taking over, and got installed with the same /etc/hosts file as Node02.

I did search the forum before, but so far noone had problem like this.
Zimbra Install doesn't even complains about the DNS error. It takes straight mydomain.net during the install on both nodes. The only problem is with starting. One host starts mydomain.net the other starts localhost even if it has the same virtual IP assigned by heartbeat

[y@w]

Your /etc/hosts file shouldn't be the same on both sides.

[SOLVED] Zimbra on DRBD

[tibby]

Your /etc/hosts file shouldn't be the same on both sides.

[SOLVED] Zimbra on DRBD

Thanks!

Can you please help me figure it out? I'm getting confused in this case this should be fine?

Node01:

127.0.0.1       mydomain.net localhost.localdomain localhost
XXX.XXX.XXX.XX3 mydomain.net Node01
XXX.XXX.XXX.XX2 Node02

Node02:

127.0.0.1       mydomain.net localhost.localdomain localhost
XXX.XXX.XXX.XX2 mydomain.net Node02
XXX.XXX.XXX.XX3 Node01

This sould work fine?

[tibby]

I've tried the config just above.

It does NOT work.
Node01 starts mydomain.net
Node02 starts localhost
Why Why Why ?

[tibby]

If I'm not wrong, you should put FQDN on the hosts file, so, your /etc/hosts should looks like this :

127.0.0.1       mydomain.net localhost.localdomain localhost
XXX.XXX.XXX.XX4 zimbrahost.mydomain.net zimbrahost
XXX.XXX.XXX.XX2 node02.mydomain.net Node02
XXX.XXX.XXX.XX3 Node01.mydomain.net Node01

Then you've saying what I've been suggested first.
To have the same hosts file on both machines.
That's not working.

Might be something outside of /opt/zimbra is configured to use localhost when starting up?

/etc/init.d/zimbra start
Host localhost
Unable to determine enabled services from ldap.

[phoenix]

Then you've saying what I've been suggested first.
To have the same hosts file on both machines.
That's not working.

Your hosts file is not the same format as the examples you've been given nor is it the same as the examples in the link I gave you earlier. Look at them again and compare it to both of your hosts files, your second host does not have an FQDN in the hosts file.

[tibby]

Phoenix:

Is this the right format?
Please correct me, what's missing here?

Node01:

127.0.0.1       mydomain.net localhost.localdomain localhost
XXX.XXX.XXX.XX3 mydomain.net Node01
XXX.XXX.XXX.XX2 Node02

Node02:

127.0.0.1       mydomain.net localhost.localdomain localhost
XXX.XXX.XXX.XX2 mydomain.net Node02
XXX.XXX.XXX.XX3 Node01

[phoenix]

Phoenix:

Is this the right format?
Please correct me, what's missing here?

Node01:

127.0.0.1       mydomain.net localhost.localdomain localhost
XXX.XXX.XXX.XX3 mydomain.net Node01
XXX.XXX.XXX.XX2 Node02

Node02:

127.0.0.1       mydomain.net localhost.localdomain localhost
XXX.XXX.XXX.XX2 mydomain.net Node02
XXX.XXX.XXX.XX3 Node01

You've already been given the answer in this thread and you do not have a correct hosts file, it doesn't match the example in the links (and how-to articles) I gave you and it doesn't match the example given to you by vavai - you're missing the FQDN for one of the nodes.

Compare your hosts file to the example in the post by vavai.