[Pkg-postgresql-private] Re: more reliable startup [was: Bug classification and cleaning request]

[Martin Pitt]

Hi!

On 2003-10-19 21:32 +0100, Oliver Elphick wrote:
> I would actually like a more reliable way of starting the postmaster. 
> pg_ctl will fork it and claim a successful start, even though that may
> not be true. 

My current idea is to check after the call to pg_ctl whether the
postmaster really runs. If yes, be quiet, if no, print a warning about
it. 

AFAICS there is not much we can do if the start really failed. When
there is a known problem (like an old or nonexisting db), there are
already warnings and an attempt to restart makes no sense. OTOH, if
there's an unknown reason why postmaster cannot start, it does not
make much sense to reattempt startup neither since we do not know what
the problem is. So the only benefit would be that the user gets a
warning message that postmaster is not currently running. I would both
print it and send a mail (because that would be a serious problem).

Any suggestions or better ideas?

Thanks,

Martin
-- 
Martin Pitt
home:  www.piware.de
eMail: martin@piware.de