Upgrade from 6.0.10NE to 7.0.0 fails with LDAP error.

[Angrysmiley]

I tried to upgrade from 6.0.10NE to 7.0.0 and the upgrade failed with the following errors:

Sat Feb 12 16:42:51 2011 Checking ldap status....
Sat Feb 12 16:42:51 2011 *** Running as zimbra user: /opt/zimbra/bin/ldap status
ldap_url and ldap_master_url cannot be the same on an ldap replica
Sat Feb 12 16:42:52 2011 not running.
Sat Feb 12 16:42:52 2011 Running bdb db_recover...
Sat Feb 12 16:42:52 2011 *** Running as zimbra user: /opt/zimbra/bdb/bin/db_recover -h /opt/zimbra/data/ldap/hdb/db
Sat Feb 12 16:42:54 2011 done.
Sat Feb 12 16:42:54 2011 Running zmldapapplyldif...
Sat Feb 12 16:42:54 2011 *** Running as zimbra user: /opt/zimbra/libexec/zmldapapplyldif
/opt/zimbra/libexec/zmldapenablereplica: line 58: [: =: unary operator expected
/opt/zimbra/libexec/zmldapenablereplica: line 293: [: =: unary operator expected
Sat Feb 12 16:43:03 2011 done.
Sat Feb 12 16:43:03 2011 Checking ldap status....
Sat Feb 12 16:43:03 2011 *** Running as zimbra user: /opt/zimbra/bin/ldap status
ldap_url and ldap_master_url cannot be the same on an ldap replica
Sat Feb 12 16:43:04 2011 not running.
Sat Feb 12 16:43:04 2011 Starting ldap...
Sat Feb 12 16:43:04 2011 *** Running as zimbra user: /opt/zimbra/bin/ldap start
ldap_url and ldap_master_url cannot be the same on an ldap replica
Sat Feb 12 16:43:05 2011 failed with exit code: 256.



Attempting to start ldap manually yields the following error:

Failed to start slapd. Attempting debug start to determine error.
config error processing olcDatabase={3}hdb,cn=config: namingContext "" already served by a preceding hdb database


Any thoughts?

The 6.0.10NE system did not have any issues prior to the upgrade

This is a single server installation on Ubuntu 8 (32bit).

[quanah]

The error you are receiving indicates some serious issue with your localconfig keys existing prior to upgrade. Line 293 references two keys:

if [ `toLower $ldap_master_host` = `toLower $zimbra_server_hostname` ]; then

line 58 references:

if [ ${ldap_port} = 636 ]; then

So, ldap_port appears to be unset, which should not be the case. The same with either ldap_master_host and/or zimbra_server_hostname

[dlbewley]

* install.sh output:

...
Restoring existing configuration file from /opt/zimbra/.saveconfig/localconfig.xml...done
Operations logged to /tmp/zmsetup.02162011-170938.log
Upgrading from 6.0.8_GA_2661 to 7.0.0_GA_3077
Stopping zimbra services...done.
Verifying /opt/zimbra/conf/my.cnf
Starting mysql...done.
This appears to be 6.0.8_GA
Checking ldap status...not running.
Running zmldapapplyldif...failed.
Checking ldap status...not running.
Starting ldap...failed with exit code: 256.
hdb_db_open: database "": db_open(/opt/zimbra/data/ldap/hdb/db/id2entry.bdb) failed: Invalid argument (22).
backend_startup_one (type=hdb, suffix=""): bi_db_open failed! (22)
bdb_db_close: database "": alock_close failed
UPGRADE FAILED - exiting.
[root@zimmer zcs-NETWORK-7.0.0_GA_3077.RHEL5_64.20110127190222]#

* /tmp/zmsetup.02162011-170938.log

[root@zimmer ~]# tail -20 /tmp/zmsetup.02162011-170938.log 
Wed Feb 16 17:10:39 2011 This appears to be 6.0.8_GA
Wed Feb 16 17:10:39 2011 Checking ldap status...
Wed Feb 16 17:10:39 2011 *** Running as zimbra user: /opt/zimbra/bin/ldap status
Wed Feb 16 17:10:40 2011 not running.
Wed Feb 16 17:10:40 2011 Running zmldapapplyldif...
Wed Feb 16 17:10:40 2011 *** Running as zimbra user: /opt/zimbra/libexec/zmldapapplyldif
IO::Socket::INET: connect: Connection refused at /opt/zimbra/libexec/zmldapapplyldif line 122.
Wed Feb 16 17:11:27 2011 failed.
Wed Feb 16 17:11:27 2011 Checking ldap status...
Wed Feb 16 17:11:27 2011 *** Running as zimbra user: /opt/zimbra/bin/ldap status
Wed Feb 16 17:11:28 2011 not running.
Wed Feb 16 17:11:28 2011 Starting ldap...
Wed Feb 16 17:11:28 2011 *** Running as zimbra user: /opt/zimbra/bin/ldap start
Failed to start slapd.  Attempting debug start to determine error.
hdb_db_open: database "": db_open(/opt/zimbra/data/ldap/hdb/db/id2entry.bdb) failed: Invalid argument (22).
backend_startup_one (type=hdb, suffix=""): bi_db_open failed! (22)
bdb_db_close: database "": alock_close failed

Wed Feb 16 17:12:08 2011 failed with exit code: 256.
Wed Feb 16 17:13:48 2011 UPGRADE FAILED - exiting.

Running again gives same effect.

I did begin by overwriting my expired eval license for 6.0.8 with a fresh eval license for 7.0.0 before running install.sh...

[dlbewley]

Looks like some db files got corrupted by a disk full condition at some point. So I just did a install.sh -u and a fresh install.

If you find yourself in a similar situation, this revision of an old post might be enlightening.

#!/bin/bash
# http://www.zimbra.com/forums/administrators/9976-ldap-will-not-start.html
#  - Uberjared
# If you're having the same problem I had, zimbra won't be running, so it
# probably shouldn't be running.  If you have slapd running...you should
# not be looking at this script.  This is a script that will resolve the problem
# I mentioned earlier in the forums.
#
# Your mileage may vary with this script.  I recreated this in the forums
# text-editor from my command-line history. Please keep that in mind.

dbd_dir=/opt/zimbra/data/ldap/hdb
cd $dbd_dir
cp -r db db.`date +"%Y%m%d"`.backup
cp -r db db.sandbox
cd db.sandbox
mkdir dumps

# Go through all .bdb files and dump them to the "dumps" subdirectory
#
for i in *.bdb ; do
  BN=`basename $i .bdb`
  /opt/zimbra/bdb/bin/db_dump -r -f dumps/$BN.txt $i
done

cd dumps

# Go through all the .txt files and load them into new .bdb files
#
for i in *.txt ; do
  BN=`basename $i .txt`
  /opt/zimbra/bdb/bin/db_dump $BN.bdb < $i
done

# Next, copy all of the .bdb files into /opt/zimbra/data/ldap/hdb/db
#
/bin/cp *.bdb $dbd_dir/db

[Angrysmiley]

Originally Posted by quanah

The error you are receiving indicates some serious issue with your localconfig keys existing prior to upgrade. Line 293 references two keys:

if [ `toLower $ldap_master_host` = `toLower $zimbra_server_hostname` ]; then

line 58 references:

if [ ${ldap_port} = 636 ]; then

So, ldap_port appears to be unset, which should not be the case. The same with either ldap_master_host and/or zimbra_server_hostname

I had a look in the localconfig, and the values are all set prior to the upgrade. Also, the server was upgraded from 6.0.9 -> 6.0.10 without any issues some time ago. And there have been no crashes/full disk etc, in between.

[dehylus]

Howto recovery this disaster!!!