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

dgamez · Post by **dgamez** » Sun Feb 13, 2011 12:33 pm

Same problem with:
"Release 7.0.0_GA_3077.RHEL4_64_20110127192526 RHEL4_64 NETWORK edition."
from: "zcs-NETWORK-6.0.7_GA_2473.RHEL4_64.20100616201318"
Also the old main.cf restore worked for me. Zimbra Forums Rocks!!

Labsy · Post by **Labsy** » Mon Feb 14, 2011 2:13 am

Huh...I obviously missed this thread prior to upgrading my production server.

Same problem here (exact as above post).
In my case senders were receiving "Relay access Denied" and problem was solved by manually issuing command under ERROR from zmconfigd.log:

zimbra@zimbraserver:/home$ cat /opt/zimbra/log/zmconfigd.log | grep ERROR
zimbra@zimbraserver:/home$ /opt/zimbra/postfix/sbin/postconf -e mail_owner='postfix' bounce_notice_recipient='postmaster' ...a lot of config here... policy_time_limit='3600' mailbox_size_limit='0' disable_dns_lookups='no'
THEN ISSUED
zimbra@zimbraserver:/home$ postfix stop

zimbra@zimbraserver:/home$ postfix start

Problem solved!

cpm · Post by **cpm** » Mon Feb 14, 2011 5:25 am

FWIW:
Upgraded zcs-NETWORK-6.0.4_GA_2038.UBUNTU8_64 to zcs-NETWORK-7.0.0_GA_3077.UBUNTU8_64
And had the same thing happen. No main.cf.
I anticipated this issue after reading this thread. Thank goodness I did.
Would have really sucked otherwise.
This needs to go in the release notes I think.
I'm also running a pcre lookup for a transport map, and my transport file didn't carry across the upgrade either. I expected that.
Surviving all this wasn't a big deal, key is in doing exactly as the docs instruct. have a full copy before you begin.

w00kie · Post by **w00kie** » Mon Feb 14, 2011 6:15 am

Long time reader - first time poster.
Platform Centos 5.5 64 Bit
Zimbra Open Source Edition
Upgrade from 6.0.10 -> 7.0
Server rejects mail with relaying denied and I have a zero byte main.cf
I tried changing permissions and running the commands provided by this thread - but I found the solution in the end was to copy the old main.cf across.
"cp /opt/zimbra/postfix-2.6.7.2z/conf/main.cf /opt/zimbra/postfix/conf"
My question is - given most of Zimbra configuration files are autogenerated - what reassurance exists that this file won't again get recreated as a zero byte file? What happens next time I add another domain or similar?
I guess it's just disappointing to see such a major bug get into a production release - Zimbra upgrades have worked flawlessly for me up until this point and considering the bug has been floating around for what seems to be a few weeks now, I'd expect the current download to have a fix.
However - its all worth it, the Carbon theme alone looks fantastic.

w00kie

phoenix · Post by **phoenix** » Mon Feb 14, 2011 6:31 am

[quote user="w00kie"]My question is - given most of Zimbra configuration files are autogenerated - what reassurance exists that this file won't again get recreated as a zero byte file? What happens next time I add another domain or similar?[/QUOTE]Nothing will happen, if you read the bug report you'll see that the cause has been found and it's been fixed. There are no guarantees in life, this is software written by humans and there will be bugs - all steps are taken to ensure it's as bug free as possible. You should, as a matter of best practice, try any upgrade on a test server before rolling it into production.
[quote user="w00kie"]I guess it's just disappointing to see such a major bug get into a production release - Zimbra upgrades have worked flawlessly for me up until this point and considering the bug has been floating around for what seems to be a few weeks now, I'd expect the current download to have a fix.[/QUOTE]Hardly a major bug (annoying for those it happens to) but it only affected a limited number of installations.

jpmtech · Post by **jpmtech** » Mon Feb 14, 2011 4:36 pm

[quote user="Labsy"]
In my case senders were receiving "Relay access Denied" and problem was solved by manually issuing command under ERROR from zmconfigd.log:

zimbra@zimbraserver:/home$ cat /opt/zimbra/log/zmconfigd.log | grep ERROR
zimbra@zimbraserver:/home$ /opt/zimbra/postfix/sbin/postconf -e mail_owner='postfix' bounce_notice_recipient='postmaster' ...a lot of config here... policy_time_limit='3600' mailbox_size_limit='0' disable_dns_lookups='no'
THEN ISSUED
zimbra@zimbraserver:/home$ postfix stop

zimbra@zimbraserver:/home$ postfix start

Problem solved![/QUOTE]
As someone that has benefited many times from the wealth of information contained in these forums I just wanted to take the time to say that this solution worked flawlessly. Thank you and thanks to all of you that make Zimbra great!

bjquinn · Post by **bjquinn** » Thu Feb 17, 2011 1:58 pm

I know this thread is turning into a "me too" fest, but I just thought I'd add that -
a) I had this same problem

b) I was upgrading from 6.0.5 -> 7.0.0 and still had the problem

c) Replacing main.cf from a backup FIXED all of my problem

d) I noticed the problem while doing the upgrade on a test server, and my typical first pass of testing is to check the output of zmcontrol status, send an email to an email address on another server, and then send an email to an address on the test server. zmcontrol status was fine, the outgoing message was fine, but the incoming message didn't work. Keep in mind this is a test server with production domains on it, so you can't send a message to it from a real email server, but you can send an email to any of the domains on the test server from the test server itself, since postfix recognizes them as domains local to the server and routes the email to the appropriate account on the test server, rather than routing it to the real production server with the same domains and accounts on it. What happened in this case was that the email went out to the REAL email account on the production server, rather than looping back to the account on the TEST server, therefore my TEST server wouldn't pass the "receiving mail" test. This was strange behavior, and was caused by the fact that since main.cf was blank, postfix didn't recognize the domains as local to the server, and therefore assumed it needed to route it externally and therefore ended up on the production server.

e) Figuring that maybe it was just the additional domains that were a problem, I sent a test message from the admin@mail.mydomain.com account to itself (mail.mydomain.com being the server name), which resulted in an error in zimbra.log of "mail for mail.mydomain.com loops back to myself".
Just thought I'd add some of my experiences in hopes that someone with my symptoms finds this thread more easily.

awaisbutt · Post by **awaisbutt** » Mon Feb 21, 2011 4:08 am

[quote user="10330phoenix"]Either one is OK.[/QUOTE]
Great work guys... I spent almost 12 hours sorting out this issue, and luckily i made a backup just before starting my upgrade which actually saved me.. For beginners, all we got to do is copy the latest folder of postfix in ur backup to the lastest postfix in the working folder. In my case I upgraded from 7.0 beta 3 to 7.0GA e.g. cp /root/Desktop/zimbra_backup/zimbra/postfix-2.6.7.2z/conf/main.cf /opt/zimbra/postfix-2.6.7.2z/conf/
ZImbra team, we need to fix this bug immediately coz this is something to do with the basic function of this mail server - SENDING EMAILS

quanah · Post by **quanah** » Tue Feb 22, 2011 2:58 pm

The bug referenced in this thread is already fixed and scheduled for release with 7.0.1.

rvasquezgt · Post by **rvasquezgt** » Sun Jun 05, 2011 11:39 pm

[quote user="quanah"]The bug referenced in this thread is already fixed and scheduled for release with 7.0.1.[/QUOTE]
It's not fixed yet, I have an Release 7.1.0_GA_3140.RHEL5_64_20110329135300 RHEL5_64 NETWORK edition.
I update from the 6.0.13, and the same error.