DX Application Performance Management

We are running CA APM 10.1 and was told that Introscope MQMoniter does not support monitoring of queue mangers hosted on multiple nodes (high availability). Can someone confirm this?

Can any APM admins, customers , and partners be the first to help our neighbor Sai with his query?

If you are referring to queue managers that are in Standy mode, then I do not think it can as it uses an MQ client connection to connect to a queue manager and then queue manager is not up.

One thing I forgot to add. We use a vip in front of the queue managers. The MQ agent is configured to point to the vip and then you don't need to worry about which queue manager is up. That is all handled by the vip.

You need to configure directly to the queue manager and not through a VIP since the agent would be routed to specific queue managers.

With the MQ & Broker agent, within the MQMonitor.properties file there is a line

mq.monitor.list=

On this line you can add additional queue managers

mq.monitor.list=QM1@host1, QM2@host1, QM3@host2

Then you duplicate the base settings

QM1@host1.host=host.company.com
QM1@host1.port=5584
QM1@host1.delaytime=180
QM1@host1.monitor.manager=minimum
QM1@host1.monitor.queue=minimum
QM1@host1.monitor.channel=minimum
QM1@host1.monitor.channelinitiator=minimum
QM1@host1.monitor.pagesets=minimum
QM1@host1.monitor.log=minimum
QM1@host1.queue.filter.exclude.system=false
QM1@host11.filter.exclude.static=true
QM1@host1.filter.exclude.dynamic = false

QM2@host1.host=host.company.com
QM2@host1.port=5594
QM2@host1.delaytime=180
QM2@host1.monitor.manager=minimum
QM2@host1.monitor.queue=minimum
QM2@host1.monitor.channel=minimum
QM2@host1.monitor.channelinitiator=minimum
QM2@host1.monitor.pagesets=minimum
QM2@host1.monitor.log=minimum
QM2@host1.queue.filter.exclude.system=false
QM2@host11.filter.exclude.static=true
QM2@host1.filter.exclude.dynamic = false

Now if you have a large number of queues/metrics from one instance, you could copy the MQ agent to another directory and then have an agent per MQ Queue Manager.

Hope this helps,

Billy

I was talking about an active queue manager and a standby queue manager which is an exact replica of the active one. It works fine to go via a vip.

We currently have our MQ properties settings setup shown below. There are 3 host (different names than the QM) instances. But whenever we have a failover, Introscope would report incorrect queue depth. Either the messages count would climb causing false positive alerts to go off. Or we don't see any activity at all. 

If we don't go through a VIP, are you suggesting that we list all 3 host ( QM1@host1, QM1@host2, QM1@host3) ?

QM1@QM1.OURDOMAIN.COM.host = QM1.OURDOMAIN.COM
QM1@QM1.OURDOMAIN.COM.port = 12345
QM1@QM1.OURDOMAIN.COM.channelname = TO.WILY
QM1@QM1.OURDOMAIN.COM.character.set = 819
QM1@QM1.OURDOMAIN.COM.ssl = disable
#QM1@QM1.OURDOMAIN.COM.channel.ssl.cipherspec = XXXXX_XXXXX
#trutstore.path = <truststore file Path>
#QM1@QM1.OURDOMAIN.COM.delaytime = 600
#QM1@QM1.OURDOMAIN.COM.report.static.freq = 20
QM1@QM1.OURDOMAIN.COM.delaytime = 60
QM1@QM1.OURDOMAIN.COM.report.static.freq = 200

## MQ Events
QM1@QM1.OURDOMAIN.COM.performance.event.queue = SYSTEM.ADMIN.PERFM.EVENT
QM1@QM1.OURDOMAIN.COM.qmgr.event.queue = SYSTEM.ADMIN.QMGR.EVENT
QM1@QM1.OURDOMAIN.COM.channel.event.queue = SYSTEM.ADMIN.CHANNEL.EVENT
QM1@QM1.OURDOMAIN.COM.event.destructive.get = true

##Metric Filters
QM1@QM1.OURDOMAIN.COM.filter.exclude.static = true
#QM1@QM1.OURDOMAIN.COM.queue.filter.exclude.system = true
QM1@QM1.OURDOMAIN.COM.queue.filter.exclude.dynamic = false
QM1@QM1.OURDOMAIN.COM.queue.filter.includeonly.regex = [ACEGNOPRSUWXZ][ABCDIMNRTWZ](?!.*ALIAS|.*RXQ).*|[CEP][LQS](?!.*ALIAS|.*RXQ).*|(.*DEAD.*)

QM1@QM1.OURDOMAIN.COM.monitor.manager = recommended
QM1@QM1.OURDOMAIN.COM.monitor.queue = recommended
QM1@QM1.OURDOMAIN.COM.monitor.channel = recommended
QM1@QM1.OURDOMAIN.COM.monitor.channelinitiator = recommended
QM1@QM1.OURDOMAIN.COM.monitor.pagesets = recommended
QM1@QM1.OURDOMAIN.COM.monitor.log = recommended

Tan.Sai,

Let us break this down into communication path and connections.  This is how I understand how the CA agent and IBM MQ functions.

There is a queue within MQ that MQ will publish performance data.  The agent will connect to the queue manager and subscribe to the queue (SYSTEM.ADMIN.PERFM.EVENT).  So the queue manager knows that if a message appears on the queue, to notify the subscriber.

Now let us say that the MQ instance fails and the backup MQ instance becomes active.  You would more than likely need to restart the agent and also initialize the performance data queue so data would be sent to the right queue on the right queue manager.

So are your fail-over hosts active hosts or are they passive hosts where they can not be communicated with till the primary fails?

If the fail-over hosts are active hosts, then yes, I am suggesting that you should list all three hosts within the configuration file for the agent.  Then within your metric groupings, expand your agent expression to capture all three possible instances.

On your queue count, you may need to build calculators to sum up the number of messages.

Hope this helps,

Billy