[SOLVED] Zimbra proxy - wrong IP for upstream server

[j2b]

I was wondering, if anybody could comment on ZCS v7 Proxy setup issues. Despite my profile info, this is connected with migration to Ubuntu server 10.04 64-bit with Zimbra v7. After installing multiserver installation (separate servers: MTA, Proxy (Zimbra), LDAP, Mailbox), Zimbra proxy seems to loose upstream server IP address or hostname. DNS resolve fine. The only thing happening similar to previous problems (solved by themselves without any changes, just letting it go for a couple of days), is when Zimbra proxy tries to connect to himself. All cli commands according to Multiserver installation was run successfully, but Zimbra proxy still wants to connect to itself, despite upstream mailbox server.

Can anybody help on solving this issue?

[Klug]

Use zmprov to change zimbraReverseProxyLookupTarget from "TRUE" to "FALSE" on the proxy server.

[j2b]

Originally Posted by Klug

Use zmprov to change zimbraReverseProxyLookupTarget from "TRUE" to "FALSE" on the proxy server.

I've rechecked - it is FALSE on proxy, and TRUE on mailbox. Still can not understand, from where zimbra takes these settings

[j2b]

If I am correct, then it turned out to be a problem of /etc/hosts and /etc/resolve.conf files.

/etc/hosts - corrected misspelling in record
/etc/resolv.conf - My DHCP server issued wrong domain and search criteria, leading to other domain rather Zimbra run on. After restart, everything works again.

Still have quesiont on nginx.log entry:

011/03/08 12:10:10 [info] 21203#0: *5 zmauth: route handler x.x.x.x:7072 sent route x.x.x.x:8080, client: y.y.y.y, server: , request: "GET /favicon.ico HTTP/1.1", host: "mail.example.com"

Why server: is empty? Does it should be so?

[j2b]

I guess this is the same type of question like [SOLVED] Spam training on multiple frontend Zimbra MTAs. If anybody will come up with specific details and spend time studying, why it should or should not be there, nobody will even drop a note for helping a little bit (zimbra people?). As far as I understand in Zimbra (especially multiserver installation), each small detail in logs can indicate potential problem, but there are vast amount of info, which in most cases do not say anything. The challenge for us, managing zimbra, is notice something, and "leave it until after something gets in trouble". And then, by spending hours, we can come up with solution, about which you were aware half year ago, to understand, that then it became a problem.

Nevertheless, main issue on Zimbra proxy was solved due to network settings. And I'd say - without running CLI scripts on Mailbox server at all - only Proxy. Although, have to say, that whether it was a problem of connection, or not fully functional setup script, but even with connection to LDAP, new server installs asked for manual correction in zmlocalconfig settings and Proxy / Mailbox port tunings.