Normally, network packet captures are only kept by the agent manager if the transaction has failed. This makes sense, because otherwise you'd need about two orders of magnitude more storage if 99% of transactions suceed.
However, we're doing some targeted triage on a complex application, and we've set up a script on the agents to remove the "human factor" out of the equation, so to speak.
Is it possible to output the packet captures somewhere for successful transactions while we're doing this test?
I'm a Little "rusty" on the SEM, but it used to be that whenever you requested a "baseline", you would capture a trace which you then would compare all subsequent traces to.
It might not be like that anymore or it might not be applicable for your case case - just a straw to grab 🙂
Another option Luke, is to grab the captures from the robots. They'll be opx or tmpopx files in the 'work' folder as usual. The names may however be meaningless (a random ID number), so if I remember your use case here right, you're only running the one application so just grab all the captures.
For scheduled transactions, I can see that the agent keeps the CNS tmpxmf files in the work folder going back about 1 hour. This should work for our current needs. However, I'm curious as to whether it's possible to get the CNS transactions if I run the script manually using the Agent Recorder, or it has to be a scheduled transaction through the Agent Manager.