Yes, it is certainly possible - assuming the standalone app is Java or .Net. You will need to define entry points in your instrumentation. OOB, we us Servlets, ASP.Net, etc. as entry points and your batch job is unlikely to use that technology. You'll need to instrument the entry point methods and set the sensor configuration to "active and start purepaths". You will also potentially need to deal with the PurePath timeout since your batch job is probably long running. You will need to instrument entry point methods that do not run for the duration of the entire job for example - you'll need to instrument sub calls. I'm sure that if you search through the forums, you'll find additional resources. Hope this helps.
The process of loading the agent into the application process is the same (e.g., for Java, edit the Java command line and add the -agentpath option).
In addition, you will need to add one or more method sensors, configured as "Active and Start PurePaths" to tell Dynatrace where transactions should begin. In general, you want this to be the method that does one unit of work. For example, if your batch job polls the database and processes any records that it finds, the right place to add a method sensor is probably the method that processes a single record.
There are three approaches to figuring this out: you can look at the source code, you can take a CPU sample and see what methods are being executed, or you can add a *lot* of method sensors. If you don't have access to the source code, I personally prefer adding a lot of method sensors (e.g., all the methods in a particular package) but you can only do that in test -- not in production.
To add to the two excellent answers. I have a YouTube Tutorial called DYnatrace for Developers. There I show how I instrument a standalone Java app. I launch the app from Eclipse and also show to to specify these entry point sensors: Dynatrace for Developers -> starting at around minute 12 I start talking about instrumenting non web apps
We run into this a lot as when dealing with COTS products. We can look into packages / classes/methods using Dynatrace's class browser but it's difficult to get working to see all or any purepaths as we see with normal JVM agent even after successful agent connection.
We can see JVM stats like GC, process CPU using infrastructure dashlets but no PurePaths.
Do you have any guidelines / recommendation to get dynaTrace working for such JVMs ?
I am running into a problems trying to instrument a standalone app that is calling out a JRE. I've managed to get the agent to connect to the right profile and tier but finding the right method to capture transactions been a little
challenging. I've tried placing sensors on, and setting the active which capture traffic but kills the application we need to monitor. Think maybe a cpu sampling might help. Do you have a youtube on cpu sampling? Some guidance please..
I believe we discussed this question via email earlier this week. I sent you these links
I do have a tutuorial on CPU Sampling and I think one on “Custom Entry
After that you sent me a CPU Sample and we identified some entry points. Hope this worked out for you