zimbraAdmin inaccessible- New install, 5.0.0 GA, RHEL5_64

[jgmora]

Installed: zcs-NETWORK-5.0.0_GA_1869.RHEL5_64.20071218220959, CentOS5 (64-bit), fresh install
No errors reported in zmsetup.01022008-105739.log, or during setup/installation.

$ zmcontrol status
Host zimbra
antispam Running
antivirus Running
ldap Running
logger Running
mailbox Running
mta Running
snmp Running
spell Running
stats Running

The basic mail interface at http://hostname/zimbra/ works as expected.

When trying to access https://hostname:7071/zimbraAdmin, the page appears to load a while, then fails abruptly without a visible error. Several service status checks from zimbramon appear in /var/log/zimbra.log, all reporting "Running" status while the page is actively loading. Various log entries related to "ProtocolHandler" appear in /opt/zimbra/log/convertd.log, but nothing that appears critical. The page behaves the same way regardless of whether iptables are loaded or not, so it does not appear to be a firewalling issue.

Any suggestions on how to troubleshoot or correct this issue (if it's a known issue?). I've attempted a complete uninstall and reinstall, but zimbraAdmin is still inaccessible.

Your assistance is appreciated.

[phoenix]

Is SElinux disabled on this machine? If it isn't then disable it and try again.

[jgmora]

SELinux is disabled via /etc/selinux/config.

It appears to be running through a series of checks and failing at some point, but I can't say for certain since nothing obvious is logged.

Anything else that should be checked?

Thank you again for your assistance.

(My original post has disappeared..?)

[phoenix]

Your original post is still there and just needed approval (anti-spam moderation), I forgot to do it last time. Could you just confirm for me that SElinux is really disabled by running 'sestatus' on the server. Do you also have any firewall on this server?

[jgmora]

[root@zimbra ~]# sestatus
SELinux status: disabled

I've tried disabling/flushing iptables and connecting from the same subnet, but it still dies after a few seconds. There is no hardware firewall between my laptop and the server.

[jgmora]

I've uninstalled and re-installed (confirming selinux was disabled) and noticed the following in the zmsetup log:

Global config attribute retrieved from ldap: zimbraSpamIsSpamAccount=
Global config attribute retrieved from ldap: zimbraSpamIsNotSpamAccount=
Global config attribute retrieved from ldap: zimbraNotebookAccount=
*** Running as zimbra user: zmlicense -c
Error: exception occurred: system failure: getDirectContext
checking isEnabled zimbra-mta
zimbra-mta is enabled
[snip]
checking isEnabled zimbra-store
zimbra-store is enabled
*** Running as zimbra user: zmlicense -c
Error: exception occurred: system failure: getDirectContext
checking isEnabled zimbra-store
zimbra-store is enabled
*** Running as zimbra user: zmlicense -c
Error: exception occurred: system failure: getDirectContext
checking isEnabled zimbra-store
zimbra-store is enabled
*** Running as zimbra user: zmlicense -c
Error: exception occurred: system failure: getDirectContext

Any other suggestions?

[jgmora]

I've reinstalled the OS and installed Zimbra from scratch, but still no go.

How to reproduce:

• Install CentOS 5 x86_64
• Disable SELinux
• Update packages
• Remove sendmail
• Disable iptables (for debugging)

• Install from zcs-NETWORK-5.0.0_GA_1869.RHEL5_64.20071218220959

Set admin password
Set license file location (/etc/ZCSLicense.xml)

Installation succeeds without any obvious error.

Go to http://hostname:7071/zimbraAdmin

After a few seconds, the page abruptly stops loading and displays no content or error messages.

**
The server has an internal (rfc1918) interface and an external interface. Is it possible that one of the zimbra services has bound to one specific address (and subsequently broken some check by the administrative interface)?
**

This configuration isn't exotic by any means, so it should work straight out of the box.

Again, thank you for your assistance.

[jgmora]

The problem appears to be caused by the multiple interfaces.

Connecting to the internal interface causes the administrative interface to load properly.

Is there a way to explicitly declare which interface the administrative interface binds to/is accessible from?

[kdsoukup]

Is it possible the URL being used is incorrect? You posted: http://hostname:7071/zimbraAdmin

But I'm pretty sure it's supposed to be https:// or did your post just contain a typo?

Sorry if it's a silly reply.

Cheers.

[phoenix]

Originally Posted by jgmora

**
The server has an internal (rfc1918) interface and an external interface. Is it possible that one of the zimbra services has bound to one specific address (and subsequently broken some check by the administrative interface)?
**

Apologies for the delayed reply. Are you saying that you have two NICs in the server? Are they both enabled and have IP addresses and are they both LAN IP addresses? If that's the case, you'll be better off disabling one of the NICs and trying the install again. If the remaining NIC is given a LAN IP address then you'll need a Split DNS set-up to get it working correctly.

There should be no problem installing Zimbra on a CentOS build, I use it myself. I do have to advise you that there is no support for installing NE on CentOS (which is an unsupported o/s). It will be OK for testing but not for ninal deployment.