Upgrade 4.5.3 to 4.5.4

[risworo]

I was trying to update my installation from 4.5.1_GA to 4.5.4_GA_763.FC5. I got to Starting LDAP ... , and it took a long time. After about 15 minutes, I checked /tmp/install.log.xxxx and foud no errors, but when I check /tmp/zmsetup.log.30832 I found this:

Checking 4.5.0_RC1
Checking 4.5.0_RC2
Checking 4.5.0_GA
Checking 4.5.1_GA
Updating from 4.5.1_GA
Getting local config tomcat_java_options
Modified tomcat_java_options=-client -XX:NewRatio=2 -Djava.awt.headless=true
Setting local config tomcat_java_options to -client -XX:NewRatio=2 -Djava.awt.headless=true
*** Running as zimbra user: /opt/zimbra/bin/zmlocalconfig -f -e tomcat_java_options='-client -XX:NewRatio=2 -Djava.awt.headless=true'
Checking 4.5.2_GA
Updating from 4.5.2_GA
Checking 4.5.3_GA
Updating from 4.5.3_GA
Global config attribute retrieved from ldap: zimbraMtaMaxMessageSize=10485760
Global config attribute retrieved from ldap: zimbraMessageCacheSize=1671168
*** Running as zimbra user: /opt/zimbra/bin/zmprov -l -- mcf zimbraMessageCacheSize 62813896
Checking 4.5.4_GA
Updating from 4.5.4_GA
Setting local config ldap_log_level to 32768
*** Running as zimbra user: /opt/zimbra/bin/zmlocalconfig -f -e ldap_log_level='32768'
Stopping ldap
Upgrade complete
Checking for port conflicts
Starting ldap...
*** Running as zimbra user: /opt/zimbra/openldap/sbin/slapindex -f /opt/zimbra/conf/slapd.conf
bdb(): Lock table is out of available locks
=> bdb_idl_insert_key: c_put id failed: Cannot allocate memory (12)

After that I hit CTRL-C and the installation went on. The rest of the zmsetup.log file did not look too worrying. However I also found this:

Session terminated, killing shell... ...killed. <--- I hit CTRL-C
*** Running as zimbra user: /opt/zimbra/bin/ldap start
Started slapd: pid 2137
*** Running as zimbra user: /opt/zimbra/libexec/zmldapapplyldif
slapd already running: pid 2137
ldap_add: Already exists (68)
ldap_add: Already exists (68)
ldap_add: Already exists (68)
ldap_add: Already exists (68)
ldap_add: Already exists (68)
ldap_add: Already exists (68)
ldap_add: Already exists (68)

After that, everything seems to run ok. However looking at /var/log/zimbra.log I found this:

<=bdb_equality_candidates: (zimbraDomainType) index_param failed (18)

The same goes for zimbraMailStatus. Does this indicate a future problem I should be aware of? If so, how would I correct it?

I am running Zimbra on an FC6 i686 machine, with 8GB of memory.

Sorry if this has been posted before. Forum search did not show related issues.
Thank you for the great product.

p.s. Then I found this [SOLVED] Messages after upgrade of 4.5.3 for 4.5.4. My stupid mistake. I apologize.

[jholder]

Hi-
I realize that you found your answer, but I needed to chime in about:

bdb(): Lock table is out of available locks
=> bdb_idl_insert_key: c_put id failed: Cannot allocate memory (12)

That actually worries me. What's to stop it from happening again? Just because things look like they're working, doesn't mean that they actually are.

So, do you have a backup/recovery strategy? Make sure you do.

john

[tbullock]

I am running Zimbra on an FC6 i686 machine, with 8GB of memory.

Is all working OK? I noticed that you installed the ZCS FC5 binary on a FC6 machine. That might be causing you some problems too.

[jholder]

GREAT catch!!

[risworo]

GREAT catch!!

So is the FC5 binary on FC6 the problem then? Before 4.5.4 I was running 4.5.1 (sorry for the mistake on the original title, should be 4.5.1 and not 4.5.3) on the same machine. I followed the instructions outlined here http://wiki.zimbra.com/index.php?tit...ora_Core_6_x86. Actually on my installation the symbolic link was already there. So did not do anything, and 4.5.1 was installed without a hitch.

Only the upgrade to 4.5.4 gave me this error:

*** Running as zimbra user: /opt/zimbra/openldap/sbin/slapindex -f /opt/zimbra/conf/slapd.conf
bdb(): Lock table is out of available locks
=> bdb_idl_insert_key: c_put id failed: Cannot allocate memory (12)

For the backup I just tar -cpf the whole /opt/zimbra directory on a spare NFS partition on another server, making sure that the permissions are kept intact.

Are there any log entries that I should watch for? Besides the following

<=bdb_equality_candidates: (zimbraDomainType) index_param failed (18)

I did not catch anything else in /var/log/zimbra.log.

Thanks.

[risworo]

So is there any log entries that I should watch for? When I installed 4.5.1 (the title on the original message was wrong) the first time, I followed the instruction for FC6 platform specific notes. Actually the symbolic link suggested on that wiki entry was already there, so I did not do anything. The installation of 4.5.1 went without a hitch.

The problem came up when I upgraded to 4.5.4 (see the quoted zmsetup.log above). Right now, the server is running, and I am testing it by transferring about 50+GB of Lotus Domino Mail. So far, no real problem. The server can be accessed through the web interface, pop3 and IMAP. So far one of my testers (100+ persons) complained.

The backup strategy is fairly simple, it involves shutting down the Zimbra processes (zmcontrol), mounting a shared NFS volumes from another server, tar -cpf to the mounted volume, disconnect the NFS, and restart the Zimbra.

Thanks.