Wondering if anyone has a clever way to enable the following behavior: If an SMF managed application fails (core, signal or all pids exit contract), do not try to restart *and* move service to a maintainance state.
The above has come up in a few situations where it is desired to have an application failure to require an administator intervention to investigate - while still having the application within the SMF framework and gaining updated service state via svcs. Making the service 'transient' is not desirable as on a process failure, the service would not be moved to maintainance (systems operations people are being groomed for svcs -x, etc). We would also like to avoid having to create a custom restarter. The only thing I can think of is to have a contract service, and create either a start or stop method script in some why to return a method $SMF_EXIT_ERR_FAIL under specific conditions (ie: under a stop->start sequence triggered by contract exit, etc). But that has implications under a 'normal' svcadm disable service; svcadm enable service event, etc. Thoughts? I know this may seem unusual to not use the actual restarter functionality to full benefit, but this has come up in a number of scenerios for adhering to existing operational procedures while transitioning to SMF. /jason
