Stork agent - application detection model (specify named, kea ip/port manually)
name: Ability to pass bind9/kea endpoints via CLI or configuration file about: Suggest an idea for this project
I wanted to run stork agent to use the nice work done on the prometheus side of this daemon. On my hardened systems (grsecurity), I tried to run on a specific unix user and runs into the following issue. Application detection model can work as process.Process() does not return "named" process.
An alternative to the whole detection mechanism is to have the ability to pass statistic channel endpoint (ip, port), eventually control channel with command line or configuration file. With that, the daemon could run on different user id or different system for example or on pod different from bind9 pod in a kubernetes deployment. Plus, this would help in unit-test. You could have multiple bind version running and check stork-agent against them.
What do you think about that ?