From nobody@hyperreal.org Mon Jul 7 11:44:39 1997 Received: (from nobody@localhost) by hyperreal.org (8.8.5/8.8.5) id LAA03145; Mon, 7 Jul 1997 11:44:39 -0700 (PDT) Message-Id: <199707071844.LAA03145@hyperreal.org> Date: Mon, 7 Jul 1997 11:44:39 -0700 (PDT) From: Alex Lau Reply-To: kclau@www.nsysu.edu.tw To: apbugs@hyperreal.org Subject: sending a kill -HUP to httpd will cause it down. X-Send-Pr-Version: 3.2 >Number: 832 >Category: general >Synopsis: sending a kill -HUP to httpd will cause it down. >Confidential: no >Severity: serious >Priority: medium >Responsible: apache >State: closed >Class: sw-bug >Submitter-Id: apache >Arrival-Date: Mon Jul 7 11:50:01 1997 >Last-Modified: Sat Sep 6 17:06:19 PDT 1997 >Originator: kclau@www.nsysu.edu.tw >Organization: >Release: 1.2.1 >Environment: OS: Solaris 2.5.1 Compiler: gcc 2.7.2.1 Except the default setting in Configuration, uncomment the: referer_log_module, expires_module, headers_module uname -a produce "SunOS www 5.5.1 Generic_103640-08 sun4u sparc SUNW,Ultra-1" >Description: command: kill -HUP `cat httpd.pid` result: bind: Invalid argument httpd: could not bind to port 80 After that, no more httpd processes exist. Restart the httpd without problem (no bind problem) >How-To-Repeat: >Fix: >Audit-Trail: State-Changed-From-To: open-analyzed State-Changed-By: marc State-Changed-When: Mon Jul 7 11:53:30 PDT 1997 State-Changed-Why: Yes, we discovered that problem under Solars the other day. It appears to be due to some obscene Solaris behavior. A temporary workaround is to add -DNO_SLACK to the EXTRA_CFLAGS line in Configuration, rerun Configure and recompile. We are thinking about the best way to deal with this. Comment-Added-By: marc Comment-Added-When: Wed Jul 9 07:39:05 PDT 1997 Comment-Added: One other workaround is to use -USR1 instead of -HUP. This does a graceful restart. See the docs for details; it is not applicable in all situations since it doesn't immediately close the logfiles, but can be used in some. Comment-Added-By: dgaudet Comment-Added-When: Sun Jul 20 10:43:46 PDT 1997 Comment-Added: We also have a patch available which fixes this bug without disabling the new slack fd feature. You can get it at http://www.apache.org/dist/patches/apply_to_1.2.1/solaris_hup.patch Dean Comment-Added-By: dgaudet Comment-Added-When: Mon Jul 28 11:47:57 PDT 1997 Comment-Added: If you applied that last patch you should reverse it (use patch -R or unpack apache_1.2.1 again and retrieve http_main.c). Then try applying the patch at: http://www.apache.org/dist/patches/apply_to_1.2.1/solaris_dup.patch (note the last one was called solaris_hup.patch). Dean State-Changed-From-To: analyzed-closed State-Changed-By: marc State-Changed-When: Sat Sep 6 17:06:19 PDT 1997 State-Changed-Why: This problem is corrected in 1.2.4. >Unformatted: