#mcollective

I am trying to automate some stuff via an mco agent/application that requires careful coordination of several systems. One step in the middle is "run puppet once". I see two choices: 1) Have the agent run puppet directly, or 2) have the applications invoke the puppet agent. Is there compelling argument for 1 vs 2?

We in some circumstances, we do need to wait for puppet to finish before proceeding.

What are the other values for "display" in the .ddl other than "always"?

i would think if you need to wait then you'd want option 1 so you can tell when the run is complete

shafer, kind of my view, but I try to be open minded. And my experience with mco is not large.

You could also rely on the status/summary actions on the puppet agent, to check if the run is done

though, calling multiple agents from one application can have some caveats; loose scoping can cause filters on agent A to "leak" over to agent B

dcorbin_: in my experience very few of the client apps pay attention to "display"