Return-Path: pg_adm@postgres.berkeley.edu Received: by postgres.Berkeley.EDU (5.61/1.29) id AA23336; Tue, 27 Jul 93 21:38:49 -0700 Date: Tue, 27 Jul 93 21:38:49 -0700 From: "David J. Hughes" Subject: Swelling backend. To: postgres@postgres.berkeley.edu Sender: pg_adm@postgres.berkeley.edu To: postgres@postgres.Berkeley.EDU Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII One piece of code I've been working on that utilises Postgres is a daemon process that hangs off a UDP socket and receives "data" from other nodes on our network. Upon receipt of a packet, it talks to Postgres for various things. Speed is important here so I leave the connection open while waiting for the next packet to turn up (could be nanoseconds or hours away). I've noticed that as time passes, the backend grows in size. I'm not sure if it's a leak or if I'm not clearing something via libpq. I have ensured that any retrieve is followed by a PQclear of the portal in use. Is there anything else that requires clearing or closing to stop this behaviour? Has anyone else noticed swelling backends if they remain open and active for a long time (my tests have generated a query rate which averages out to about one every 3 seconds sustained for half an hour or so (although you can see growth within a few quick bursts of 25 queries)). Any clues? The backend eventually dies due to lack of memory so I'd like to resolve this without PQreset'ing and forcing a new backend (and the resultant performance hit). ___ David J. Hughes bambi@Bond.edu.au / \ / / / / __/ __ __ ____/ / / __ Senior Network Programmer / \ / \ / \ / / / / / \ / Information Technology Services \____/ \__// / \__/ \___/ / / / Qld. 4229 AUSTRALIA (+61 75 951450)