[SOLVED] Where does Zimbra Proxy get mailbox upstream IP address.

[j2b]

I am troubleshooting inconsistent IMAP functionality in multiserver installation, and wanted to find out, where does Zimbra Proxy get IP address for exact mailbox server. We deploy multiserver environment and split DNS. Mailbox server on mailbox, LDAP and Proxy resolve to DMZ IP address, but nginx.log displays upstream IP address, which is NATted. The connections go through, but can not find resource, where NGINX gets this IP. Have to troubleshoot firewall too.

[j2b]

Somehow it was solved, even can not explain. As there are no comments from anybody during the month, I guess, nobody knows, even from Zimbra, where is it taken from.

To be sure, please check once more your DNS and network settings, especially if you use DHCP in your configuration, as I did.

[dalmate]

it gets from ldap's database.

[Klug]

The IP addresses are not in the LDAP database.

zimbra-proxy asks the mailboxes servers that are setup as "zimbraReverseProxyLookupTarget".
It asks the LDAP to know which server have this parameter set to TRUE then direct connects to them.

Do you need your DNS to be correct, because it relies on the DNS...

[j2b]

Due to fact, that my issue took place at the beginning of this year, and up to my knowledge and understanding somehow solved without any specific signs to me on possible configuration changes, I'd say that DNS and Server restart were actions, which led to status normalization.

Initial issue: All settings for proxy operations were set according to wiki manuals, although it were available for v6, but no for v7. Servers initially were restarted, and the only place to find any inconsistency was nginx.log, where it was saying, that no resources for destination server were found, despite correct DNS setup and doubled records in /etc/hosts file. E.g. Proxy server didn't have a clue, where to turn to for mailbox, leaving destination IP whether to proxy server or leaving IP address blank. But after some time of experiments and a couple of system restarts, everything started to work. So no additional help I can give more, nor my notes on this issue are kept any more

Additionally we were struggling with timeouts and Java errors copying large mailboxes, but not all of them. Thus we had to solve these issues with non transferrable email messages in another way - appologizing to our customers. Which for sure, is not a solution in technical terms