[SOLVED] Upgrade Problem

[iiibill]

My upgrade is a long tail of woe that is all my own fault. I upgraded to Debian lenny before I saw that this was a problem and I tried to upgrade to 5.0.18 zcs on the lenny system. This pretty much porked the configuration. I have reinstalled the OS, now it is etch again, and have most of the configuration back.

What I have hit is an ldap authentication error near the end of the install.sh run.

ERROR: service.FAILURE (system failure: unable to lookup server by name: pobox.ncsd.k12.ca.us message: [LDAP: error code 49 - Invalid Credentials]) (cause: javax.naming.AuthenticationException [LDAP: error code 49 - Invalid Credentials])

I have tried resetting the password with zmldappasswd and that did not fix the problem. Any suggestions would be greatly appreciated.

Bill

[phoenix]

The error message says that it can't find the server, it does a DNS lookup for that - so can it be resolved? Post the output of the following commands (run on the zimbra server):

cat /etc/hosts
cat /etc/resolv.conf
dig yourdomain.com mx
dig yourdomain.com any
host `hostname` <-- use that exact command with backticks not single quotes

[iiibill]

Arrrgh. Thanks for the sharp eyes. Thought I had resolved those problems. Here's the output. Note, I did not include the mx lookup because this configuration never had one. The mail routing, content scanning, and virus scanning are done on another server. We use zimbra as a mailbox server only at this point.

Note, the terminology is always a bit confusing to me. I don't normally think of pobox.ncsd.k12.ca.us as a domain name, it is a host name. But, in Zimbra-ese I think that it is a domain name.

Bill

/etc/hosts
osts
127.0.0.1 localhost
204.89.147.9 pobox.ncsd.k12.ca.us
204.89.150.5 ldapauth.ncsd.k12.ca.us

/etc/resolv.conf
search ncsd.k12.ca.us
nameserver 204.89.147.1

pobox:/opt/package/zimba/zcs-5.0.18_GA_3011.DEBIAN4.0.20090707170635# dig pobox.ncsd.k12.ca.us any

; <<>> DiG 9.3.4-P1.1 <<>> pobox.ncsd.k12.ca.us any
;; global options: printcmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 4906
;; flags: qr aa rd ra; QUERY: 1, ANSWER: 1, AUTHORITY: 3, ADDITIONAL: 3

;; QUESTION SECTION:
;pobox.ncsd.k12.ca.us. IN ANY

;; ANSWER SECTION:
pobox.ncsd.k12.ca.us. 3600 IN A 204.89.147.9

;; AUTHORITY SECTION:
ncsd.k12.ca.us. 604800 IN NS dns2.ncsd.k12.ca.us.
ncsd.k12.ca.us. 604800 IN NS ns1.macallister.grass-valley.ca.us.
ncsd.k12.ca.us. 604800 IN NS dns1.ncsd.k12.ca.us.

;; ADDITIONAL SECTION:
ns1.macallister.grass-valley.ca.us. 604800 IN A 72.249.145.228
dns1.ncsd.k12.ca.us. 3600 IN A 204.89.147.1
dns2.ncsd.k12.ca.us. 3600 IN A 204.89.150.5

;; Query time: 1 msec
;; SERVER: 204.89.147.1#53(204.89.147.1)
;; WHEN: Mon Jul 13 10:48:16 2009
;; MSG SIZE rcvd: 183

[iiibill]

That 'osts' is a cut and paste error. Here is the correct hosts file.

127.0.0.1 localhost
204.89.147.9 pobox.ncsd.k12.ca.us
204.89.150.5 ldapauth.ncsd.k12.ca.us

[iiibill]

So, I tried just renaming /opt/zimbra out of the way and creating a new zimbra installed. In the new install zimbra starts up just fine. That says to me that hosts, hostname, and the DNS are fine. It is something broken in the production installation that I can figure out.

[iiibill]

Oh, and just for completeness here is the host lookup results.

pobox:/opt# host `hostname`
pobox.ncsd.k12.ca.us has address 204.89.147.9

[iiibill]

Okay, I finally read the ldap trouble shooting page. I started the ldap server manually. then set the login level to 256 and ran:

ldapsearch -x -h pobox.ncsd.k12.ca.us -D "cn=config" -W

and got to see everything in the directory. Did this from pobox and from another system. The ldap server seems to be running just fine. Zimbra is not finding it. As I said I don't think this is a DNS/resolver issue. There is some blinking setting that I am missing. Still digging, but if someone else had a clue I would be grateful.

[iiibill]

I guess this is now my personal blog. I have moved on to a new error after manually starting slapd and setting the uid=zimbra,cn=admins,cn=zimbra userpassword to match what is in zmlocalconfig. Now the install.sh upgrade goes to the end and the last message is:
... lots of output ...
Mon Jul 13 19:12:38 2009 Starting mysql
Mon Jul 13 19:12:50 2009 Starting logger mysql
I can't upgrade version

UPGRADE FAILED - exiting

At least it is a new error.

[iiibill]

The root problem appears to have been that the failed upgrade replaced every
password in the zimbra configuration with a new value. To get this system running
again I had to do the following.

1. Start up the mysql server with --skip-grant-tables and set the root and zimbra passwords to match zmlocalconfig.
   
2. Start up the mysql logger server in a similar manner and reset root and zimbra passwords
   
3. Reset the ldap admin password. I did this with zmldappasswd.
   
4. Reset the postfix password in the ldap directory. I did this using ldapmodify using the dn=config credentials with a simple bind.
   
5. Reset the uid=zimbra,cn=admins,cn=zimbra password in the ldap directory. I did this using ldapmodify using the cn=config credentials with a simple bind.
   
6. Recreate the ca and the cert for the jetty server. I followed the instructions on the wiki.