From: Tom Lane <tgl@sss.pgh.pa.us>
To: Merlin Moncure <mmoncure@gmail.com>
cc: Mack Talcott <mack.talcott@gmail.com>,
 postgres performance list <pgsql-performance@postgresql.org>
Subject: Re: Debugging shared memory issues on CentOS
In-reply-to: 
 <CAHyXU0wv+ONa_ou+0bH92EcmX2smfRyh_ikN-vLvXe87p6GiWw@mail.gmail.com>
References: 
 <CAPZbZCDrS11_YAExC_tHfv7U=1Phzbqi3=kTkp6Kd7CM4M0RBg@mail.gmail.com>
 <3716.1386737676@sss.pgh.pa.us>
 <CAPZbZCDEGY9nPtFOrXqYynt4oOk9RRsSiR+G_4BUNQmK5QK6pA@mail.gmail.com>
 <22204.1386819597@sss.pgh.pa.us>
 <CAHyXU0wv+ONa_ou+0bH92EcmX2smfRyh_ikN-vLvXe87p6GiWw@mail.gmail.com>
Comments: In-reply-to Merlin Moncure <mmoncure@gmail.com>
 message dated "Thu, 12 Dec 2013 08:56:59 -0600"
Date: Thu, 12 Dec 2013 10:17:08 -0500
Message-ID: <3931.1386861428@sss.pgh.pa.us>
Precedence: bulk
Sender: pgsql-performance-owner@postgresql.org

Merlin Moncure <mmoncure@gmail.com> writes:
> It's tempting to say, "there should be a limit to backend local cache"
> but it's not clear if the extra tracking is really worth it all things
> considered.  There was some discussion about this (see the archives).

Yeah --- there actually was a limit on total catcache size once, long ago.
We took it out because it was (a) expensive to enforce and (b) either
pointless or counterproductive on most workloads.  The catcache is
probably the least of the memory hogs anyway, so it might be that limiting
the size of relcache or function caches would be more useful.  But that
memory is likely to discourage most hackers from investigating.

			regards, tom lane


-- 
Sent via pgsql-performance mailing list (pgsql-performance@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-performance