We’re using 2 servers in production so that we have redundancy for our infobright servers. Things have been working quite well for about a week. Then yesterday on our passive master during a load we got this error in the infobright.log file:

2009-07-06 14:38:42 Internal error: An object unlocked too many times in memory manager.
2009-07-07 09:49:19 Error: Could not open or lock required tables.

During the Load the script got this error from this server:
  “Lost connection to MySQL server during query”

I was not able to shut down infobright with the /etc/init.d/mysql-ib stop command. I tried twice and got this error msg in the bh.err file the first time:
90707 9:49:19 [Note] SCHEDULER: Purging queue. 0 events
090707 9:49:21 [Warning] /usr/local/infobright-3.1.1-x86_64/bin/mysqld: Forcing close of thread 6100 user: ‘myuserblah’

But no other errors while I tried to get infobright to shut down. I did the ole kil -9 on the pid. Then I started infobright & stopped it in the normal manner & infobright started & stopped normally.

The server appears to be working fine now.

The primary server loading the same files did not have a problem.

DETAILS:
Master DB server uname -a
Linux primary-db 2.6.24-23-generic #1 SMP Mon Jan 26 01:04:16 UTC 2009 x86_64 GNU/Linux
Secondary DB server uname -a (problem server)
Linux secondary-db 2.6.24-23-generic #1 SMP Wed Apr 1 21:43:24 UTC 2009 x86_64 GNU/Linux

Both are running:
Server version     5.1.14-log
infobright-3.1.1-x86_64

The only difference is one is running infobright from a symlink (primary server):
/usr/local/infobright  

The secondary server is running infobright from
/usr/local/infobright-3.1.1-x86_64

Guess I’ll check on the infobright.ini & my-ib.cnf differences.

Thoughts? Shoud I file a bug report?

thanks,
erin

Hardware differences:
primary-db = 24GB RAM
secondary-db = 32GB RAM

The only real differences are in the brighthouse.ini file:
< ServerMainHeapSize=24000
—-
> ServerMainHeapSize=10000
27c27
< ServerCompressedHeapSize=4000
—-
> ServerCompressedHeapSize=1000
35c35
< # ControlMessages = 0
—-
>  ControlMessages = 2

Questions:
1. Should I set the HeapSizes down? and to what numbers???
2. Should I turn on ConrolMessages? (it’s set to # ControlMessages = 0 )

thanks,
erin

Hi Erin

What is the relation between master, slave ICE servers?

It’s not clear how they co-operate with each other. I assume you are not using mysql replication rather using some external replication which is not aware to the ice servers. I also assume you are just running mysqld servers from same directory but each has separate instance as well as separate data folder. If so, then I would say, this error should not be related master/slave architecture but rather is an internal defect.

We have defect similar to “...An object unlocked too many times in memory manager.”. Please also note “ICE replication” has some limitation and also not well tested.

best regards,

I call it old school replication.

We just take the data files & load them onto two servers. No MySQL replication. They are TWO separate servers that have no idea about each other. But we have HA with 2 servers that have the same data so if one server fails or we need to do maintenance we can switch to the other. Its my understanding that this is the recommended form of “replication” for infobright.

We need to be sure that both servers have all data files loaded & that they are both ready for action. our “passive” server (meaning we just load data on to it but all queries are happening on the “active” server) failed. Lucky for us it was the passive. The error was strange. I have no answers to the OPs folks on why this server failed. It helped them to figure out how to set up Nagios to check for this error in the future. But if we need to make changes to the app I’ll need more details.

Sorry if I wasn’t clear about our form of HA.

erin