Zimbra fails to start after installation

[Alex--]

Hi all.

I've just downloaded Zimbra Network Edition (trial) and attempted installation on my test box which is CentOS 4.2 machine (equivalent to RHEL4 Update 2).

During installation, I got some error messages, and Zimbra failed to start. The error messages were just after "initializing LDAP" and certificate creation:

Creating server entry for t115.test.8-p.biz...*** Running as zimbra user: /opt/zimbra/bin/zmprov cs t115.test.8-p.biz
Done
Setting spell check URL to http://t115.test.8-p.biz:7780/aspell.php...
*** Running as zimbra user: /opt/zimbra/bin/zmprov ms t115.test.8-p.biz zimbraSpellCheckURL http://t115.test.8-p.biz:7780/aspell.php
Done
Setting service ports on t115.test.8-p.biz...
*** Running as zimbra user: /opt/zimbra/bin/zmprov ms t115.test.8-p.biz zimbraImapBindPort 143 zimbraImapSSLBindPort 993 zimbraPop3BindPort 110 zimbraPop3SSLBindPort 995
Done
Adding t115.test.8-p.biz to zimbraMailHostPool in default COS
ERROR: service.FAILURE (system failure: getDirectContext) (cause: javax.naming.CommunicationException t115.test.8-p.biz:389)
ERROR: service.FAILURE (system failure: getDirectContext) (cause: javax.naming.CommunicationException t115.test.8-p.biz:389)
Done

Similary, if I try to start Zimbra from command line, I'm getting following:

t115# /etc/init.d/zimbra restart
Host t115.test.8-p.biz
        Stopping antispam...Done
        Stopping antivirus...Done
        Stopping ldap...Done
        Stopping logger...Done
        Stopping mailbox...Done
        Stopping mta...Done
        Stopping snmp...Done
        Stopping spell...Done
Host t115.test.8-p.biz
        Starting ldap...Done.
ERROR: service.FAILURE (system failure: getDirectContext) (cause: javax.naming.CommunicationException t115.test.8-p.biz:389)
t115#

Doing the netstat -an, shows nothing is listening on port 389.

I did default installation. The only things I changed were domain (from t115.test.8-p.biz to test.8-p.biz) and set admin password for LDAP.

I'm kinda stuck. Any way to get this thing going?

[goetzi]

The only things I changed were domain (from t115.test.8-p.biz to test.8-p.biz) and set admin password for LDAP.

So you reinstalled it? But normally, this shouldn't be a problem as i know.

I have Network Edition running on CentOS 4.2 Server too and also had this problems.

I think it's your /etc/hosts file. What do you have in there?

It should look like this (maybe it's working with another hosts too, but for me it is working):

127.0.0.1          localhost.localdomain localhost
192.168.0.15     host.domain.com host

At the second line, just insert the IP-Adress, FQDN and hostname of your machine.

Then do remove Zimbra (./install.sh -u) and install it again.

[Alex--]

After posting, I found couple of threads with similar problem. Many mentioned "do this or that to /etc/hosts". In my case, nothing helped. I even went step further and created A record for domain (test.8-p.biz) pointing to the host where I was installing Zimbra (I had MX record pointing there too). Hasn't made any difference.

My original /etc/hosts was the standard one. The way /etc/hosts file should look like:

127.0.0.1              localhost.localdomain   localhost
192.168.111.115        t115.test.8-p.biz       t115

I've attempted removing t115 line, merging it with localhost line, and whatnot (all the stuff from other threads). Nothing helped.

[goetzi]

And your DNS configuration is working? Can t115.test.8-p.biz be resolved?

[Alex--]

Yes, it can be resolved. test.8-p.biz is resolvable only from internal network (not from Internet). I've checked with dig, returns both MX record (pointing to t115.test.8-p.biz) and A record (IP address 192.168.111.115). A record for t115 is correct too (192.168.111.115).

[marcmac]

Does ldap start? Run ldd /opt/zimbra/openldap/libexec/slapd and look for missing libraries.

Is the firewall running?

Does localhost resolve to 127.0.0.1 (host localhost)

[Alex--]

No firewall. localhost is resolvable from both local files and DNS (nsswitch.conf says "hosts: files dns", so it gets it from /etc/hosts).

All libraries needed for ldap are present:

t115# ldd /opt/zimbra/openldap/libexec/slapd
        libsasl2.so.2 => /opt/zimbra/cyrus-sasl/lib/libsasl2.so.2 (0x00e6d000)
        libssl.so.4 => /lib/libssl.so.4 (0x00cb7000)
        libcrypto.so.4 => /lib/libcrypto.so.4 (0x002cf000)
        libresolv.so.2 => /lib/libresolv.so.2 (0x00d8f000)
        libpthread.so.0 => /lib/tls/libpthread.so.0 (0x00111000)
        libc.so.6 => /lib/tls/libc.so.6 (0x007ea000)
        libdl.so.2 => /lib/libdl.so.2 (0x00b50000)
        libgssapi_krb5.so.2 => /usr/lib/libgssapi_krb5.so.2 (0x00e20000)
        libkrb5.so.3 => /usr/lib/libkrb5.so.3 (0x00123000)
        libcom_err.so.2 => /lib/libcom_err.so.2 (0x00b23000)
        libk5crypto.so.3 => /usr/lib/libk5crypto.so.3 (0x0019b000)
        libz.so.1 => /usr/lib/libz.so.1 (0x00d6c000)
        /lib/ld-linux.so.2 (0x0022f000)
t115#

[Alex--]

BTW, LDAP service does not start. The only service running after installation is MySQL. After doing "/etc/init.d/zimbra restart" no Zimbra services are running.

[marcmac]

Not everyone pays attention to the nsswitch.conf file when deciding how to resolve hostnames. I'm specifically referring to ldap and postfix - which is why I asked for the output of `host localhost`. If that returns 127.0.0.1, you're good.

So, if ldap starts (su - zimbra, ldap start) can you telnet to port 389?

ANything in /tmp/install.log or /tmp/zmsetup.log?

[Alex--]

OK, found it...

I was installing on "minimal installation" CentOS 4.2 box. The installation manuall for Zimbra says "acceppt all default answers" (which means full Desktop install, are they out of their mind?)... So, after I did full Desktop install, then installed Zimbra, it started up just fine.