where to configure queues & channels name for their monitoring through Prometheus on cloud paks for integration

[sumitkantsinha]

Hello,

i have created a queue manager using cloud paks for integration V2020.1.1 over Openshift 4.4, i am looking after the monitoring capabilities using bundled Prometheus & Grafana. I could see some queue manager stats coming as default in the grafana dashboard, but can't find how to configure the queue names or channel names to monitor. I could see the one of the K8 services object is created for port 9157, but can't see any MQ service objects created which listen on port 9157 on container . Need help to find how to configure queues & channels monitoring here & where is the monitoring service running on mq container.

ocpadmin@linuxvm:~/ocp-install$ ./oc rsh -n mq mqhelmtest-ibm-mq-0 sh-4.4$ dspmq QMNAME(mqhelmtest) STATUS(Running) sh-4.4$ sh-4.4$ runmqsc mqhelmtest 5724-H72 (C) Copyright IBM Corp. 1994, 2019. Starting MQSC for queue manager mqhelmtest. AMQ8521I: Command completion and history unavailable.

dis service() 1 : dis service() AMQ8629I: Display service information details. SERVICE(SYSTEM.AMQP.SERVICE) AMQ8629I: Display service information details. SERVICE(SYSTEM.DEFAULT.SERVICE)

   :

end 2 : end One MQSC command read. No commands have a syntax error. All valid MQSC commands were processed. sh-4.4$ sh-4.4$ pwd / sh-4.4$ sh-4.4$ ls @System.solv bin boot dev etc home lib lib64 licenses lost+found media mnt opt proc root run sbin srv sys tmp usr var sh-4.4$ ls opt MQOpenTracing mqm sh-4.4$ ls opt/mqm READMES bin gskit8 inc instinfo.tsk java lib lib64 licenses mqpatch.dat msg samp swidtag web sh-4.4$ sh-4.4$

Thanks, Sumit

[arthurbarr]

The metrics endpoint only offers queue manager-level metrics. If you want queue- or channel-specific metrics, I'm afraid you'll need to add your own metrics endpoint. This can be done in addition to the main metrics, using the mq-metric-samples. You can run another Pod which will subscribe to the queue manager pub/sub topics, and emit them as Promtheus metrics.

[sumitkantsinha]

Hi Arthur,

Thanks for your reply.

so, where is the metric services hosted in the container for which we have port 9157 listening it ?

Also i am of the view that by giving options for queue & channel specific metrics using Prometheus without any external/ other pod would also eliminate the need of any custom monitoring tool / scripts & at the same time if the feature is bundled within, it would give me at least more option to convince my customer on cost reduction for running their MQ on cloud paks for integration.

would like to get your thoughts on it...

[arthurbarr]

The current metrics are hosted by the runmqserver process, which creates a local bindings connection to the queue manager to listen for message on the pub/sub statistics topics.

I agree that it is a good idea to add queue and channel specific metrics, but it just hasn't been done yet. It isn't entirely trivial, as you can't just subscribe to "all queues", and have to be told names of specific queues.