Zimbra Proxy Upgrade Error From 6.0.8 to 7.0

[det]

Hi,

After I upgrade a multi-server install of Zimbra 6.0.8 to 7.0, I encountered this when running zimbra in proxy server. All other servers (ldap,mta,mailbox) upgraded successfully. The proxy server also upgraded but it does not start imapproxy and nginx due to below error. Looks like it added an extra "conf" on the path.


Starting nginx...nginx: [emerg] open() "/opt/zimbra/conf/conf/nginx/includes/nginx.conf.main" failed (2: No such file or directory) in /opt/zimbra/conf/nginx.conf:5

Thanks.

[sikanen@leappoint.fi]

Same issue here :S

[momohteks]

Hi there,

I'm facing the same issue upgrading from 6.0.12 to 7.1.2.

Did you find any fix for that issue ?

Thanks.

[basictheprogram]

Same problem here. Any fix or work-around?

[momohteks]

Yes,

Fixed by upgrading first to 7.1.1, and then upgrade to 7.1.2

Hope that helps

[ghantoos]

FYI, upgrading from 6.0.14 to 7.1.1 does not work. After installing the new packages, the mysql service fails to start, hence preventing the install from proceeding.

Cheers,
Ignace M

[apg]

I had the same problem and did the following.

created sim link from . to conf

# cd /opt/zimbra/conf
# ln -s . conf

Now upon restarting (zmcontrol stop/start) have this error:
Starting nginx...nginx: [emerg] open() "/opt/zimbra/nginx-0.9-zimbra/log/nginx.log" failed (2: No such file or directory) failed.

So then:

created sim link for log directory:
# cd /opt/zimbra/nginx-0.9-zimbra
# ln -s logs log

That seems to have fixed the problem...

[apg]

Ok... just got off the phone with support. What happened is that I had virtual hosts that I had since deleted from our DNS servers. So during the upgrade zmproxyconfgen failed leaving the configs in /opt/zimbra/conf/nginx/includes incorrect; along with /opt/zimbra/conf/nginx.conf. When we manually ran /opt/zimbra/bin/zmproxyconfgen it complained about the missing IP:
Exception in thread "main" com.zimbra.common.service.ServiceException: system failure: Cannot find the IP of xxxx.yyyy.zzz.edu

The fix for that was using the management interface, going to domains, and then the virtual hosts tab and removing the unused virtual host(s). If they were actually being used we could've added them to our DNS servers.

Apparently zmproxyconfgen reads the config from ldap, and then uses the templates in /opt/zimbra/conf/nginx/templates/ to generate the config files in /opt/zimbra/conf/nginx/includes

Here's the applicable bug reports:
Bug 66072 – Non resolvable zimbraVirtualHostname prevents upgrade from 6.x.x to 7.x.x
Bug 54481 – Can't find included nginx conf file after upgrading to nginx-0.9-zimbra