Hari Babu <haribabu(dot)kommi(at)huawei(dot)com> writes:
> I have a query regarding the changing of pmState only incase of SIGTERM but
> not incase of SIGINT.
> Is there any reason to handle like this? or Is it ok to modify the handling
> of pmState in SIGTERM same as SIGINT?
AFAICT the "smart shutdown" code path is okay: it transitions to
PM_WAIT_READONLY state, and then PostmasterStateMachine issues SIGTERM
to the startup process and goes to PM_WAIT_BACKENDS state. Perhaps this
could be done more clearly but I'm disinclined to mess with it.
regards, tom lane