[SOLVED] Zimbra 6 -> 7 - Zimbra does not know its own virtual hosts

pnunn · Post by **pnunn** » Tue Feb 08, 2011 5:05 pm

Thankyou... fixed my install too... the power of backups

holunde · Post by **holunde** » Tue Feb 08, 2011 5:35 pm

Well, I can confirm most of the above.

I'm running Zimbra in a production-environment with small ressources and only about 200 users. Without a test-system I'm forced to do the upgrades at night and hope for the best.

Upgrading from 6.0.10 to 7 broke the server as described above, but I couldn't nail down the error, so I had to move back to 6.0.10 using rsync.

Tonight I tried again with help of all your feedback, and indeed the problem is the postfix main.cf file being empty after upgrading. Using the old main.cf

with 644 zimbra:zimbra permissions solved the problem.

Otherwise the 6->7 upgrade was extremely smooth, and I applause the Zimbra guys for that.
But still, this seems to be a major bug that could harm an otherwise correctly configured system. How did this not get caught in the test-phase?

cwall64 · Post by **cwall64** » Thu Feb 10, 2011 4:09 pm

Interestingly enough, I had two servers on 6.0.10 (and Ubuntu 8.04). I performed the Ubuntu upgrade first, as suggested in these forums, then the Zimbra upgrade, and one server (most used/taxed one) had a blank main.cf, the other was fine... Running the failed postconf command from the logs solved my issue.

liveaverage · Post by **liveaverage** » Fri Feb 11, 2011 8:13 am

Just wanted to note that I did not experience this issue upgrading from 6.0.8 to 7.0 on Ubuntu 8.04 LTS 64-bit. I did make a quick copy of my main.cf just-in-case, but the permissions and contents of the file were well-maintained through the upgrade process. I had this thread bookmarked in the event of a problem

ekkas · Post by **ekkas** » Fri Feb 11, 2011 2:54 pm

So which is the best workaround?
A)"If you hit the issue with postfix, do *not* delete main.cf. Instead, I suggest examining the zmconfigd.log file and running the postconf command it logs from the command line as a workaround."
or
B) Copy main.cf from backup

phoenix · Post by **phoenix** » Sat Feb 12, 2011 1:20 am

[quote user="ekkas"]So which is the best workaround?[/QUOTE]Either one is OK.

fyd · Post by **fyd** » Sat Feb 12, 2011 7:10 am

[quote user="liveaverage"] I had this thread bookmarked in the event of a problem

[/QUOTE]
Yeah me too.

nice work.
I am guessing if this is hit only with upgrades from 6.0.10 to 7.0. Because I read some others had safe upgrade from 6.0.9, and most of them who reported were using 6.0.10.

phoenix · Post by **phoenix** » Sat Feb 12, 2011 7:28 am

[quote user="fyd"]I am guessing if this is hit only with upgrades from 6.0.10 to 7.0.[/QUOTE]It only affects upgrades to ZCS 7 and it doesn't affect everyone.

fyd · Post by **fyd** » Sat Feb 12, 2011 8:05 am

[quote user="10330phoenix"]It only affects upgrades to ZCS 7 and it doesn't affect everyone.[/QUOTE]
Yeah, I meant 'only with 6.0.10 to 7.0' and not with 6.0.8/6.0.9 to 7.0. Not sure though, just a guess after reading some.

MNPF · Post by **MNPF** » Sat Feb 12, 2011 1:29 pm

Had the same problem on CentOS release 5.5 (Final) 32bit upgrading from 6.0.10_GA_2692.RHEL5_20110127201734 CentOS5 FOSS edition 32bit to 7.0.0_GA_3077.RHEL5_20110127201734 CentOS5 FOSS edition 32 bit.
Did the workarround of main.cf and its working fine now.