Thread: ldap error
View Single Post
  #15 (permalink)  
Old 11-18-2007, 09:54 PM
PhishKiller PhishKiller is offline
Active Member
 
Posts: 48
Default

After fixing my /etc/hosts file and rebooting the system.. Zimbra refused to work again.

After a lot more googling around.. I took the advice that I needed to be SURE that all processes were killed. Sadly.. the /etc/init.d/zimbra stop/start script is horriblhy inadequate. When slapd starts quit... the script did not indiated anything had gone wrong. Nor, did it tell me the one time I went right.

If there is not a but to change the start/stop/restart script to be more UNIX like (e.g. show [OK] [FAILED] messages for EVERY step of start/stop.. here.. let me show an example of properly crafter start/stop script incase the team that wrote the zimbra one has never seen one. Observer the status message for each component of the system's shutdown and resart. Simple.. but VERY effective when managing a system.

Code:
-bash-3.1# /etc/init.d/sendmail restart
Shutting down sm-client:                                   [  OK  ]
Shutting down sendmail:                                    [  OK  ]
Starting sendmail:                                         [  OK  ]
Starting sm-client:                                        [  OK  ]
-bash-3.1#
So.. let me close my rant with this. After making sure I killed everything zimbra related ( ps -ealf | grep zimbra ) I uninstalled it for the 10th or so time.. and ran one more install.

That time it worked. Properly. Now I have been able to turn the system over to my mail admin for final configuration. It's takem more than a week to get this up and running.

I looked at your /etc/hosts.. looks OK to me. Check your MX and hostname digs and see if those are all pointing at the same server.

One of the initial 'mistakes' we made was using a generic name of 'mail' for our server.. but we already had a 'mail' MX record for our domain. What we did was create an MX record for 'testmail.ourdomain.com' which pointed to the server named 'mail.ourdomain.com' (which was resolvable via DNS and also /etc/hosts). The problem this caused was the startup script sometimes used the DNS, sometimes used the MX entry and in other cases based it's automated $url paramters (in the setup scripts), making for some very frustrating installation messages.

My advice:

- kill all zimbra processes
- make sure the MX, DNS and /etc/hosts ALL reference the same system and IP.
- uninstall zimbra (install.sh -u)
- re-instal zimbra (install.sh)
- keep fingers crossed.
- let us know if it worked.
Reply With Quote