I have a websphere cell with ~20 clusters and 3 nodes, so ~20 servers on each node.
Now, I have the same amount of different agent groups, each for every cluster. The agent groups config are basicly the same, or with very few different. But sometimes I have to put some custom sensors on a cluster, what is not necessary for the other ones. It's a little bit difficult to manage so many agent groups. I am thinking about to have only one agent group for every was clusters.
Do you think is a good idea to change?
What is your configuration recommendation for this kind of architecture?
Thanks
Answer by Andreas G. · Jan 24, 2013 at 05:51 PM
The idea with agent groups is that you basically describe a logical tier with one agent group. If you then have 10,20 or even more servers that are part of the same tier in a clustered environment you would still use the same agent group as they logically are the same tier.
It is not wise to create so many agent groups - just for the reasons you mentioned -> too hard to maintain.
So - best practice is to define an agent group by logical tier. Every agent group will also show up as a Tier-node in the Transacton Flow with the option to collapse and expand this tier to actually see each individual server that is part of that tier.
Andi
Learn the enhanced capabilities of the next generation Dynatrace AI root cause analysis and how to feed it with your own data sources.
Wednesday, February 20, 2019
Register today!
Learn the enhanced capabilities of the next generation Dynatrace AI root cause analysis and how to feed it with your own data sources.
Wednesday, February 20, 2019
Register today!
Would you like to have an early taste of what we have cooked up for 2019? We would love to hear your feedback and improve some of the new features. Check NAM 2019 Beta release notes.
Sign up today!