a - no, I don’t have a RA at hand, but have toyed with many of 'em, including self-written ones.
b - you might want to take a look at the agent script and determine what it’s trying to do. Might be as simple as an too early initial check for a functional instance after bringing up the resource.
If we were to assist with debugging, more detailed information is in order. I recommend that you take a closer look at the startup messages and try to determine what’s going wrong. The usual way things operate would be
- the desired state in the CIB is set to “started”
- the LRM invokes the RA to start the resource (does that work? You might want to manually set up a temp environment and invoke the RA to see what’s actually happening)
- after the RA reports back success, LRM will periodically invoke the RA to check the resource status
So basically you should check if startup did indeed succeed - or were there errors, probably due to some missing or wrong parameters or because startup took longer than the granted time? If startup seemed ok, did the RA report rc=0 when checking the resource state? If not, why (probably checked to early?)
If you still don’t find your way around this (no-one is going to blame you for that ) you should provide the resource definition and probably related syslog output.