This is my architecture:
1. A deployment server (DS) with apps within deployment-apps folder
2. Two search heads (SH)
3. Two clustered indexers (CI)
4. A number of production servers with Splunk forwarder installed (SF).
I feed both SH and SF with apps on DS. SF write to CI and SH reads from CI.
I need to develop a custom app to read application logs. I know that I need inputs.conf on SF and props.conf (with i.e. EXTRACT parameter set) on SH.
What is the best and proper way to do so? Create two separate apps (one for SH and one for SF)? Or use one app, but disable part of config files somehow (i.e. basing on server class)? I don't want inputs.conf to be used on SH and props.conf on SF. What about Splunkbase apps (i.e. for Apache) - how they should be different and how to maintain them?