Process group detection rules, extracting the identifier

[kalle_lahtinen]

Hi,

I have some WAS processes which are named in a way that's difficult for Dynatrace to properly group them. I'm using the attribute "WebSphere server" for this. There's first a version number in the middle, and then in the end a node number. So e.g.:

NAME_01_APP_01

The NAME is a constant for all processes, and I'm using that as the starting delimiter ("NAME_"). So from this, I'd like to extract "01_APP" as the process group. My question is, if I define the end delimiter as _0, does it then pick the first _0 or the last _0? So which direction is it reading the string from, from the left or from right?

Since there's no preview function available and these only take place after a restart, it's not so straightforward to test... if someone has the info, I'd appreciate it 🙂

[Julius_Loman · Accepted solution]

Actually the best method here is to utilize the DT_CLUSTER_ID and DT_NODE_ID environment variables for those WebSphere instances and not dealing with the process group detection rules at all.

Just set your desired values for the server process in WAS admin console in :

Application servers > <server > Process definition > Environment Entries:

And fully restart the WAS process.
From the example above, your WAS will be a member of MyProcessGroup and will be identified as node1.

[kalle_lahtinen]

Thanks Julius! That does indeed look like the correct way to do it. But for clarity's (and documentation's) sake, I still would like to understand how that PG detection rule works in the example I presented; how is that "end delimiter" evaluated in case the substring appears more than once within the attribute? I think that's good to understand for other cases, anyway.

[kalle_lahtinen]

Hey,

Having now tested this. I can confirm that the end delimiter is being read from the right hand side of the string. So my PG detection rule in this case worked without having to edit the env. variables; NAME_01_APP_01 and NAME_01_APP_02 went into processes 01_APP (Node01) and 01_APP (Node02).

[Julius_Loman]

Glad it worked for you.

Process group rules is really an area to improve. For example I'm really missing the regex functionality here (extract the process group name and node id by using capture groups).