MQSeries.net :: View topic

darring · Posted: Wed Sep 10, 2008 1:28 pm Post subject:

Hi,
We are facing some performance issues with Aggregate nodes.
We are running MQSI 6.0 on Solaris. While going through the documentation,
I realized that we need to create MQSI_AGGR_MQ environment variable and
populate it with broker name and the execution group name to force the
broker to use MQ rather than Database table for aggregation. But befor doing
that I tried to take a look at the SYSTEM.BROKER.AGGR.* queues,
interesting thing is I can see that all the execution groups using Aggregate
nodes are using those queue, where as we don't have the MQSI_AGGR_MQ environment
variable defined in the broker user profile.

So question is do we really need MQSI_AGGR_MQ to force broker to use MQ rather than Databse to control Aggregate?

If we must need the MQSI_AGGR_MQ environment variable, according to documentation MQSI_AGGR_MQ needs to be set as
MQSI_AGGR_MQ=<Broker List comma separated values>/<Execution Group Name List comma separated values>. Now lets consider the scenario,
I have brokers B1 and B2, Broker B1 having execution groups Egp1, Egp2 and broker B2 having execution groups Egp1, Egp3 and I want
the Egp1 of B1 and Egp3 of B2 to use MQ and Egp1 of B2 to carry on with Database to control Aggregate nodes, I will be specifying
the environment variable as below
MQSI_AGGR_MQ=B1,B2/Egp1,Egp3
But looking at the MQSI_AGGR_MQ environment variable value, broker doesn't have any way to identify
that I want the Egp1 of Broker B2 to carry on with Databse to control Aggregate. Isn't it funny????

Thanks