[SOLVED] Upgrade issues from 5.0.18GA to 6.0.8 Debian 5

[gebbeth]

Ok, so I had been running 5.0.18GA on debian 4 with perl downgraded. I initially went to install and had some perl issues, but after futzing around with Debian, it settled on 5.10 in Lenny and the install went through without any more perl issues.

I then ran into issues of the upgrade not being able to find ldap.bak. The install was resetting it to a 0 byte file. I found a post that said to set the perms to 440 on the file and the upgrade proceeded with only 1 other error: "Upgrading Document templates...failed." I also had some port conflicts but I had Zimbra set up to listen on alternate ports in the old install so I didn't think to much of it.

The upgrade completed and all of the services show started when I do a zmcontrol status. I can't however get either the web based UI or the admin UI to load. The pages just time out. I see port 7071 listening in lsof output but I do not see the https (port 444) listening. When I telnet to the mail port, I get a mail prompt, so I suspect it is accepting and queuing mail.

Update:
After looking at the mailbox.log file, I find the following:
"2010-10-26 12:01:55,821 WARN [main] [] misc - Could not establish a connection to the database. Retrying in 5 seconds.
com.zimbra.common.service.ServiceException: system failure: getting database connection
ExceptionId:main:1288112515820:63e54b52c9e60a62"

I am not even sure where to start troubleshooting database errors.

[phoenix]

Have a look at these threads first: site:zimbra.com +"Could not establish a connection to the database" - Yahoo! Search Results

[gebbeth]

I looked at these threads and they all have to do with msql that won't load or isn't running. My mysql is running. Nmap shows that port 7306 is open. lsof shows that the port is listening and is owned by mysqld.

In the case where mysql is running, what would cause it not to connect? Is zimbra trying to connect to mysql on a different port perhaps? How would I check that?

101027 13:40:41 mysqld started
101027 13:40:41 InnoDB: Started; log sequence number 0 377729457
101027 13:40:41 [Note] /opt/zimbra/mysql/libexec/mysqld: ready for connections.
Version: '5.0.90-log' socket: '/opt/zimbra/db/mysql.sock' port: 7306 Source distribution

[gebbeth]

Stranger and stranger. Using tcpdump listening on port 7306 on the lo (localhost) interface, I see the beginnings of a tcp connection handshake, a tcp packet with the SYN bit set destined for localhost:7306. Then I get a reply back from localhost with the RST and ACK bits set which resets the connection. Every 5 seconds the process repeats itself as another connection attempt is made. The packet capture is attached in libpcap format.

14:01:44.373852 IP localhost.localdomain.57426 > localhost.localdomain.7306: S 1243514974:1243514974(0) win 32792 <mss 16396,sackOK,timestamp 34174524 0,nop,wscale 6>
14:01:44.373887 IP localhost.localdomain.7306 > localhost.localdomain.57426: R 0:0(0) ack 1243514975 win 0
14:01:44.374993 IP6 localhost.55066 > localhost.7306: S 953617556:953617556(0) win 32752 <mss 16376,sackOK,timestamp 34174525 0,nop,wscale 6>
14:01:44.375012 IP6 localhost.7306 > localhost.55066: R 0:0(0) ack 953617557 win 0
14:01:44.375656 IP localhost.localdomain.57428 > localhost.localdomain.7306: S 1240217792:1240217792(0) win 32792 <mss 16396,sackOK,timestamp 34174525 0,nop,wscale 6>
14:01:44.375684 IP localhost.localdomain.7306 > localhost.localdomain.57428: R 0:0(0) ack 1240217793 win 0
14:01:44.376662 IP6 localhost.55068 > localhost.7306: S 956881478:956881478(0) win 32752 <mss 16376,sackOK,timestamp 34174525 0,nop,wscale 6>
14:01:44.376676 IP6 localhost.7306 > localhost.55068: R 0:0(0) ack 956881479 win 0
14:01:46.377137 IP localhost.localdomain.57430 > localhost.localdomain.7306: S 1270747135:1270747135(0) win 32792 <mss 16396,sackOK,timestamp 34175025 0,nop,wscale 6>

[gebbeth]

Edited the my.cnf to have bind-address = 127.0.0.1 instead of the device's hostname.