Received: from hapcibeno.freestart.hu (hapcibeno.freestart.hu [213.197.64.8]) by postgresql.org (Postfix) with ESMTP id 317D147545D for ; Sat, 20 Apr 2002 04:47:32 -0400 (EDT) Received: from line-68-160.dial.freestart.hu ([213.197.68.160] helo=jaguar) by freestart.hu with smtp (Freestart relay 2 (FS-Lin)) id 16yqeH-00014E-00; Sat, 20 Apr 2002 10:55:13 +0200 Message-ID: <008301c1e848$4e5b9a40$01fdfea9@jaguar> Reply-To: "Papp, Gyozo" From: "Papp, Gyozo" To: , , References: <1463.192.168.1.205.1019057195.squirrel@www.saolucas-se.com.br> <200204181017.56709.jmpoure@translationforge.com> <011901c1e6b6$e990c5a0$01fdfea9@jaguar> <200204200916.28425.jmpoure@translationforge.com> Subject: Re: PERFORMANCE Date: Sat, 20 Apr 2002 10:49:38 +0200 MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-2" Content-Transfer-Encoding: quoted-printable X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 6.00.2600.0000 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000 X-Scanner: exiscan by Freestart X-Archive-Number: 200204/72 X-Sequence-Number: 1096 Hi, Paulo wrote: | I=B4m new on the list and I=B4ve been using postgres on a production=20 | environment for 3 months. My team and I have developed a system for a | Hospital in which we work with php-postgres-linux. In the begining=20 | everything was fine. As soon as the tables grew larger, we=B4ve begun to | experience an enormous performance fall. By now, the system is and Jean-Michel wrote: | I figured out the only way to "crash" PostgreSQL was entering an endless = loop.=20 | This happened to me sereval times (a trigger was spreading to another tri= gger,=20 | which in turned called back the first trigger; or simply an endless loop = in a=20 | PLpgSQL function). When entering an endless loop, PostgreSQL activity cli= mbs=20 | up, untill it reaches 100% user activity and dramaticaly slows down Linux= .=20 | Isn't what happens to you? Turn on debuging to notice where the endless l= oop=20 | happens. Yes, Jean-Michel is right, please try to debug what queries sent to the bac= kend. BTW, I'm doubt that it would be an endless loop rather than mis configured = or misconcepted (if you understand what I mean) database design. Paulo stated there were times when everything was fine, pg just recently sl= owed down. It might be a very evil trigger which starts to misbehave as tables g= row. BTW, it could be, really. If that trigger works too slowly...