From nobody@hyperreal.com Wed Apr 23 15:46:01 1997 Received: (from nobody@localhost) by hyperreal.com (8.8.4/8.8.4) id PAA04694; Wed, 23 Apr 1997 15:46:01 -0700 (PDT) Message-Id: <199704232246.PAA04694@hyperreal.com> Date: Wed, 23 Apr 1997 15:46:01 -0700 (PDT) From: Dean Moeller Reply-To: dmoeller@bell-labs.com To: apbugs@hyperreal.com Subject: after Broken pipe, server gets into a hard loop X-Send-Pr-Version: 3.2 >Number: 466 >Category: os-unixware >Synopsis: after Broken pipe, server gets into a hard loop >Confidential: no >Severity: critical >Priority: medium >Responsible: apache >State: closed >Class: sw-bug >Submitter-Id: apache >Arrival-Date: Wed Apr 23 15:50:00 1997 >Last-Modified: Wed May 14 09:22:14 PDT 1997 >Originator: dmoeller@bell-labs.com >Organization: >Release: 1.2b6 >Environment: Unixware 1.1 >Description: This has happened quite often: The server quits responding to a browser and floods the error log with entries. While I haven't been able to reproduce this at will, it seems to happen when I am on a "frames" browser page and attempt to link somewhere else. Anyway, attached is a copy of the error_log and access_log at a time when things get goofy: ERROR_LOG: [Mon Apr 21 16:03:34 1997] Server configured -- resuming normal operations ******* MY COMMENTS HERE ********************** (normal stuff here) Now the following happens until I SIGHUP server or error log fills.... *********************************************** [Wed Apr 23 15:20:32 1997] accept: Broken pipe [Wed Apr 23 15:20:32 1997] - socket error: accept failed [Wed Apr 23 15:20:32 1997] accept: Connection reset by peer [Wed Apr 23 15:20:32 1997] - socket error: accept failed [Wed Apr 23 15:20:32 1997] getsockname: Invalid argument [Wed Apr 23 15:20:32 1997] accept: Connection reset by peer [Wed Apr 23 15:20:32 1997] - socket error: accept failed [Wed Apr 23 15:20:32 1997] getsockname: Invalid argument [Wed Apr 23 15:20:32 1997] accept: Connection reset by peer [Wed Apr 23 15:20:32 1997] - socket error: accept failed [Wed Apr 23 15:20:32 1997] getsockname: Invalid argument [Wed Apr 23 15:20:32 1997] accept: Connection reset by peer [Wed Apr 23 15:20:32 1997] - socket error: accept failed [Wed Apr 23 15:20:32 1997] getsockname: Invalid argument [Wed Apr 23 15:20:32 1997] getsockname: Invalid argument [Wed Apr 23 15:20:32 1997] accept: Connection reset by peer [Wed Apr 23 15:20:32 1997] - socket error: accept failed (lots deleted) [Wed Apr 23 15:25:11 1997] getsockname: Invalid argument [Wed Apr 23 15:25:11 1997] accept: Connection reset by peer [Wed Apr 23 15:25:11 1997] - socket error: accept failed [Wed Apr 23 15:25:11 1997] getsockname: Invalid argument [Wed Apr 23 15:25:11 1997] getsockname: Invalid argument [Wed Apr 23 15:25:11 1997] accept: Connection reset by peer [Wed Apr 23 15:25:11 1997] - socket error: accept failed [Wed Apr 23 15:25:11 1997] accept: Connection reset by peer [Wed Apr 23 15:25:11 1997] - socket error: accept failed [Wed Apr 23 15:25:11 1997] getsockname: Invalid argument [Wed Apr 23 15:25:11 1997] accept: Connection reset by peer [Wed Apr 23 15:25:11 1997] - socket error: accept failed [Wed Apr 23 15:25:11 1997] getsockname: Invalid argument [Wed Apr 23 15:25:11 1997] accept: Connection reset by peer [Wed Apr 23 15:25:11 1997] - socket error: accept failed [Wed Apr 23 15:25:11 1997] getsockname: Invalid argument [Wed Apr 23 15:25:11 1997] accept: Connection reset by peer [Wed Apr 23 15:25:11 1997] - socket error: accept failed [Wed Apr 23 15:25:11 1997] getsockname: Invalid argument [Wed Apr 23 15:25:11 1997] SIGHUP received. Attempting to restart [Wed Apr 23 15:25:12 1997] Server configured -- resuming normal operations ACCESS_LOG: projrm2 - - [23/Apr/1997:15:16:43 -0600] "GET /cgi-bin/logs?ua+0 HTTP/1.0" 200 2204 projrm2 - - [23/Apr/1997:15:16:45 -0600] "GET /cgi-bin/logs?ua+1 HTTP/1.0" 200 114 projrm2 - - [23/Apr/1997:15:17:27 -0600] "GET /cgi-bin/logs?smtp+0 HTTP/1.0" 200 164233 projrm2 - - [23/Apr/1997:15:17:40 -0600] "GET /cgi-bin/logs?ua+0 HTTP/1.0" 200 114 projrm2 - - [23/Apr/1997:15:18:05 -0600] "GET /cgi-bin/logs?smtpsched+0 HTTP/1.0" 200 118 projrm2 - - [23/Apr/1997:15:18:08 -0600] "GET /cgi-bin/logs?smtp+0 HTTP/1.0" 200 164233 projrm2 - - [23/Apr/1997:15:18:21 -0600] "GET /cgi-bin/logs?smtpd+0 HTTP/1.0" 200 115121 projrm2 - - [23/Apr/1997:15:18:55 -0600] "GET /cgi-bin/logs?smtp+0 HTTP/1.0" 200 164233 projrm2 - - [23/Apr/1997:15:30:27 -0600] "GET /cgi-bin/iim HTTP/1.0" 200 1098 dradap - - [23/Apr/1997:16:12:47 -0600] "GET /cgi-bin/iim HTTP/1.0" 200 1098 >How-To-Repeat: I'm behind a firewall >Fix: >Audit-Trail: State-Changed-From-To: open-analyzed State-Changed-By: coar@decus.org State-Changed-When: Thu Apr 24 08:09:54 PDT 1997 State-Changed-Why: Thank you for this report. We are looking into methods of addressing this. State-Changed-From-To: analyzed-open State-Changed-By: dgaudet State-Changed-When: Thu May 1 14:29:16 PDT 1997 State-Changed-Why: User reports that the unixware docs didn't help, and neither does 1.2b10. I'm not a unixware user, so I'll just reopen this one. Dean State-Changed-From-To: open-feedback State-Changed-By: coar State-Changed-When: Tue May 13 12:00:42 PDT 1997 State-Changed-Why: I believe this has been fixed out-of-band by defining USE_FCNTL_SERIALIZE_ACCEPT and disabling UW's cc optimisation in in the server build. Can you please confirm that this problem is closed? Category-Changed-From-To: general-os-unixware Category-Changed-By: coar Category-Changed-When: Tue May 13 12:00:42 PDT 1997 State-Changed-From-To: feedback-closed State-Changed-By: coar State-Changed-When: Wed May 14 09:22:13 PDT 1997 State-Changed-Why: Feedback from customer: >Unformatted: >I have built with USE_FCNTL_SERIALIZE_ACCEPT and no optimization and >haven't seen the problem again. If the problem returns, I'll resend >a trouble report. Thanks to everyone who helped me. > >Dean Moeller Thanks for using Apache!