From nobody@hyperreal.com Thu May 29 20:26:01 1997 Received: (from nobody@localhost) by hyperreal.com (8.8.5/8.8.5) id UAA13253; Thu, 29 May 1997 20:26:01 -0700 (PDT) Message-Id: <199705300326.UAA13253@hyperreal.com> Date: Thu, 29 May 1997 20:26:01 -0700 (PDT) From: Brian Tiemann Reply-To: btman@pacific.net To: apbugs@hyperreal.com Subject: CGI scripts sometimes don't run, but most of the time they're fine X-Send-Pr-Version: 3.2 >Number: 631 >Category: os-bsdi >Synopsis: CGI scripts sometimes don't run, but most of the time they're fine >Confidential: no >Severity: serious >Priority: medium >Responsible: apache >State: closed >Class: support >Submitter-Id: apache >Arrival-Date: Thu May 29 20:30:01 1997 >Last-Modified: Sun Jun 22 21:44:04 PDT 1997 >Originator: btman@pacific.net >Organization: >Release: 1.2b11 (all) >Environment: BSD/OS www.pacific.net 2.1 BSDI BSD/OS 2.1 Kernel #1: Mon May 5 10:17:03 PDT 19 97 dpm@www.pacific.net:/usr/src/sys/compile/PACIFIC i386 >Description: There are times, seemingly unrelated to the load on the machine, when any and all output from perl scripts just doesn't work. If it's an entire page output from a CGI, the result is a 500 error. If it's an SSI, it returns [an error occurred while processing this directive]. Most of the time, all scripts run with no problem. Server load is generally about 0.3, and the problem goes away immediately if I reload Apache. Could this be a problem with the number of spare servers running? Min is 5, Max is 15. Each of these glitches seems to last from 2 to 5 minutes. >How-To-Repeat: Try looking at http://www.lionking.org for the SSI's, and http://www.lionking.org/today.cgi for the 500 error. >Fix: >Audit-Trail: State-Changed-From-To: open-analyzed State-Changed-By: dgaudet State-Changed-When: Sun Jun 22 21:36:31 PDT 1997 State-Changed-Why: There could be a file descriptor leak somewhere, maybe something caused by the bsdi c library. If you can catch the server while it's doing this, then try running lsof against some of the children and the parent. They should all have about the same number of open files +/- two or three. Are there any errors in the error_log when this happens? Dean State-Changed-From-To: analyzed-closed State-Changed-By: dgaudet State-Changed-When: Sun Jun 22 21:44:03 PDT 1997 State-Changed-Why: User reports the problem went away with 1.2 final. >Unformatted: