[SOLVED] Message: system failure: getting database logger connection on admin login

[captainmish]

Hello

ubuntu 606
zimbra NE 5.0.5

Ive seen a couple of forum posts about similar errors, but they all seem to say that the logger service is dead. This is not the case here, as zmcontrol says that all processes are running fine:

zmcontrol status
Host mail.changeddomain.com
antispam Running
antivirus Running
ldap Running
logger Running
mailbox Running
mta Running
snmp Running
spell Running
stats Running

ps shows 10 "/usr/bin/perl /opt/zimbra/libexec/zmlogger" processes running as well.

Yet I still get the error when logging in to the admin gui:
Server error encountered Message: system failure: getting database logger connection Error code: service.FAILURE Method: GetServiceStatusRequest Details:soap:Receiver

I have tried stopping and starting the logger with zmloggerctl stop/start, but this error continues on each login, Ive tried a myisamchk under /opt/zimbra/logger/db/data/zimbra_logger, also get same results.

Anyone have any ideas, am I missing something?

[uxbod]

Are you able to connect to the database via

su - zimbra
logmysql

also take a look in /opt/zimbra/logger/db/data for a file named <zmhostname>.err. See if there are any problems being identified in it. Also, have a check through your Log Files - Zimbra :: Wiki for any error messages.

[captainmish]

thanks uxbod - some useful info there!
I tried logmysql and it errored out with "ERROR 1040 (08004): Too many connections"

I have upped the max_connections in /opt/zimbra/conf/my.cnf, then restarted the logger with zmloggerctl - I could connect suggesfully with logmysql after this, but still get the error. I am not sure if the new config is being read, is there a better way to get zimbra's mysql to re-read its config?

Cheers,

[uxbod]

su - zimbra
zmcontrol stop ; zmcontrol start

that will read everything then.

[captainmish]

It was actually /opt/zimbra/conf/my.logger.cnf that I wanted. I upped the max connections to 20 and restarted the logger, and it all seems happy. Should I be worried that it hit max connections?

[uxbod]

You could also set it via

su - zimbra
zmlocalconfig -e zimbra_mysql_connector_maxActive=20

with respect to anything to worry about, are you seeing any other strange error messages ? How many users are connected to the system ? Or multiple people using the Admin GUI ?

[captainmish]

Ah, should have known zmlocalconfig could do it, thats useful - I guess it will stop my edit being broken on upgrades.
There are about 100 users on this server, and possibly one more admin - cant seem to find anything hugely worrying in the logs, but i have not done an exhaustive search of them all.

[captainmish]

Originally Posted by uxbod

You could also set it via

su - zimbra
zmlocalconfig -e zimbra_mysql_connector_maxActive=20

with respect to anything to worry about, are you seeing any other strange error messages ? How many users are connected to the system ? Or multiple people using the Admin GUI ?

This wont work for this particular problem, because the logger gets its config from ${zimbra_home}/conf/my.logger.cnf, not the my.cnf used by other mysql stuff. (Set via zmlocalconfig, then logmysql and show variables, max connections still at default 10) There doesnt seem to be a localconfig setting for the logger mysql max connections:

zmlocalconfig | grep mysql
logger_mysql_bind_address = localhost
logger_mysql_data_directory = ${zimbra_home}/logger/db/data
logger_mysql_directory = ${zimbra_home}/logger/mysql
logger_mysql_mycnf = ${zimbra_home}/conf/my.logger.cnf
logger_mysql_pidfile = ${zimbra_home}/logger/db/mysql.pid
logger_mysql_port = 7307
logger_mysql_socket = ${zimbra_home}/logger/db/mysql.sock
[snip]

So its a config file edit and remember on each upgrade for now..

[qhartman]

Thanks for posting this. I ran into this same problem with my Zimbra installation this morning and figuring out the problem on my own would have taken quite some time!

Like the OP, I am a little concerned about what the root cause of this might be. We've only got about 80 users on the system, and very few use the webmail. I am the only admin.

[Rich Graves]

Did any of you file a bug or support case?

Do you have a large number of volumes or CPUs (as I do) that might explain this?

Ran into this again at

zmlogger dying under 5.0.11