<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta http-equiv="Content-Type" content="text/html;charset=ISO-8859-1">
<title></title>
</head>
<body>
Tom Lane wrote:<br>
<blockquote type="cite" cite="mid25853(dot)1032724594(at)sss(dot)pgh(dot)pa(dot)us">
<pre wrap="">fredrik chabot <a class="moz-txt-link-rfc2396E" href="mailto:fredrik(at)f6(dot)nl"><fredrik(at)f6(dot)nl></a> writes:<br></pre>
<blockquote type="cite">
<blockquote type="cite">
<pre wrap="">What version of Postgres? What platform (Linux, etc)? <br></pre>
</blockquote>
</blockquote>
<blockquote type="cite">
<pre wrap="">psql (PostgreSQL) 7.2.1<br>SMP Linux 2.2.18pre22<br></pre>
</blockquote>
<pre wrap=""><!---->I seem to recall that SMP support was very flaky in the 2.2 series<br>kernels, and even in 2.4 up to 2.4.10 or something like that.<br>Perhaps a ker<pre
wrap="">nel update would make life better.</pre></pre>
</blockquote>
I turned on loggin and it appeard to be a shared memory locking problem,
so I migrated the database to a non-smp machine running RH 7.2 and the postgres
7.2.1-5 rpm's<br>
<br>
Kernel 2.4.7-10<br>
<br>
and I still get
<blockquote type="cite" cite="mid25853(dot)1032724594(at)sss(dot)pgh(dot)pa(dot)us">
<pre wrap="">FATAL 1: LWLockAcquire: can't wait without a PROC structure</pre>
</blockquote>
The code tells me this should never heppen :-<<br>
<br>
It is less frequent than on the smp machine.<br>
<br>
regards fredrik chabot<br>
</body>
</html>