Re: [HACKERS] Postmaster dies with many child processes - Mailing list pgsql-hackers

From Cary O'Brien
Subject Re: [HACKERS] Postmaster dies with many child processes
Date
Msg-id 199901301408.JAA16054@saltmine.radix.net
Whole thread Raw
List pgsql-hackers
Tatsuo suggested...

> Or even better, MaxBackendId can be set at the run time such as
> postmaster's option. Also, it would be nice if we could monitor number
> of backends currently running. Maybe we should have a new protocol for
> this kind of puropose?

Oracle does this with pseudo-tables.  There are about (from my memory)
15 tables, all starting with v$, that let you look into the internals
of a running oracle database.  Things like mounted tablespaces, locks,
all sorts of things.  One is (I think) just a name,value table for
misc parameters.  There were even things like writes/second and
reads/second for performance tuning.

We did a bunch of nice tk front ends for monitoring oracle simply
by looking at the v$ tables.  

Using a tables like this (pg_config, pg_backends, pg_transactions,
whatever) would be one way to provide this information.

I haven't delved into backend code enough to know if this
is possible.

-- cary





pgsql-hackers by date:

Previous
From: Ian Grant
Date:
Subject: ERROR: heap_fetch: xinv19201 relation: ReadBuffer(81aeefe) failed
Next
From: "Patrick Welche"
Date:
Subject: Re: [HACKERS] new heap manager mmalloc