MTA service error after power cut

[russgalleywood]

Hi Guys,

after the helpful Electricity Company man turned off the power without warning at a small company I help, we are getting errors with their Zimbra installation!
And, before you ask, no I hadn't gotten around to adding a UPS to the setup yet, so I know it's my own fault for not doing so!

But on rebooting the Zimbra server they are not getting any local mail or appointment requests.

I'm pretty sure it's something to do with the MTA service but a zmcontrol status or stop and start does not throw up any errors.
I think perhaps it's a old process before the power cut still running that needs shutting down.
We are on CentOS 5 and Zimbra 5, I am not at the building of the company so am not sure on the exact version.
Any ideas Guys?
Thanks
Russ

[Bill Brock]

First thing you probably should do is scour the logs. The answer may jump out at you.

Next, do you have a backup?

What file system are you running on your server? Is it journalizing?

[russgalleywood]

Thanks Bill,

I have had a look at the logs but could not see much amiss, the only thing was in the latest mta log file which mentioned the MTA process already active when stopping and starting Zimbra.
I am certainly no expert though as you can probably tell!

I do have backups on the Zimbra server of course, would the easiest thing be to just restore the backup from before the power cut? It would not lose much mail as we have also had broadband issues since that day, BT acidentally cut us off!

It is running on ext3 which is not journalising I believe.

Thanks

Russ

[russgalleywood]

Straneg thing is, we are definitely sending mail out through a POP account and receiving in through the same account so it must be something about our local accounts since the power cut. It was cetainly working fine before then.

[Bill Brock]

Without log entries showing any errors I'm afraid I am at a loss as well. I hate to tell you to do a restore and that not fix the problem. But that is one reason why we do backups.

Maybe someone else will chime in with an idea.

I think Ext3 does journalize. Which is a great file system feature. I use reiserfs on all my Linux boxes and have been very pleased on the rare occasion my UPS gave out before power is restored. It's very reassuring to watch those transaction being replayed.

[russgalleywood]

Thanks Bill.

I will post logs but am not sure which one I shoul;d be posting.

I'm sure it's something really simple but just can't figure it out!

It's not even showing any deferred messages in the mail queue.
I have tried sending using the basic client too but no difference.

I am downloading 5.0.5 at the moment to upgrade from 5.0.4 to see if this cures it.

Thanks for your help once again.

Russ

[William]

Originally Posted by russgalleywood

Thanks Bill,

I have had a look at the logs but could not see much amiss, the only thing was in the latest mta log file which mentioned the MTA process already active when stopping and starting Zimbra.

Russ

Netstat show any processes hogging the ports?

[russgalleywood]

Thanks for helping.
After stopping Zimbra there is not much else running, nothing I can see that should cause a clash.
But as I said before, I'm no expert!

I am going to upgrade to 5.0.5 after backing up.
If that doesn't work I guess a restore is the best thing.
What do you think Guys?

Thanks
Russ

[uxbod]

With ZCS shutdown run

netstat -an

to see whether anything else is holding ports open. Then once you start ZCS again check /var/log/zimbra.log as that should show you any Postfix errors. What does

su - zimbra
zmcontrol status

show when you have restarted ZCS ?

You can also check /opt/zimbra/logger/db/data/.err for any MySQL errors aswell. If there are any errors you could try a autorepair of the tables with

su - zimbra
/opt/zimbra/libexec/zmdbintegrityreport -r

or to do it from within MySQL

su - zimbra
logmysql
use zimbra_logger;
show tables; <- this will give you a list of all the ones to check, but the error file should say which one you need to check/fix
check table mta <- or replace mta with the one(s) from the error file
repair table mta <- again replace mta with the one(s) from the error file
quit
zmcontrol stop ; zmcontrol start

[russgalleywood]

Thanks for your helpful reply Uxbod, especially at the weekend!

Netstat -an doesn't show anything unusual and on stopping and starting the services zmcontrol status says all are working correctly.

There are no errors in SQL but the zimbra.log for yesterday does show some MX errors, I will investigate further.

Thanks

Russ