* Tom Lane <tgl@sss.pgh.pa.us> [001129 20:22]:
> > I'd lean towards a pg_ping (Peter E., any comment here?)
>
> > Really we'd need to change the postmaster too, because what we need to
> > do is send a query "are you ready to accept connections?" that the
> > postmaster will answer without an authentication exchange. AFAIR this
> > is *not* immediately evident from the postmaster's current behavior ---
> > I think it will challenge you for a password even before the startup
> > subprocess is done.
>
> I fixed that today; if the database status is not open-for-business,
> the postmaster will tell you so right away instead of making you go
> through the authentication protocol first. So a pg_ping could be
> written that just sends a connection request packet and sees what
> comes back.
>
> However, if we're running in TRUST or IDENT mode, it's possible that
> that technique will lead to launching a backend to no purpose. So
> maybe we ought to extend the postmaster protocol to have a "query
> status" packet type. Thoughts?
I'd also like to see a protocol extension or some such to maybe
collect SNMP or other statistical data that could be used later for
tuning. If we do a protocol change, let's make it extensible....
LER
>
> regards, tom lane
--
Larry Rosenman http://www.lerctr.org/~ler
Phone: +1 972-414-9812 E-Mail: ler@lerctr.org
US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749
