From nobody@hyperreal.com  Mon Apr 21 22:22:16 1997
Received: (from nobody@localhost)
          by hyperreal.com (8.8.4/8.8.4)
	  id WAA18523; Mon, 21 Apr 1997 22:22:16 -0700 (PDT)
Message-Id: <199704220522.WAA18523@hyperreal.com>
Date: Mon, 21 Apr 1997 22:22:16 -0700 (PDT)
From: Josh Rabinowitz <joshr@stockmaster.com>
Reply-To: joshr@stockmaster.com
To: apbugs@hyperreal.com
Subject: The 'root' httpd server process will block, the machine will respond very slowly, and refuse all connections.
X-Send-Pr-Version: 3.2

>Number:         448
>Category:       os-sunos
>Synopsis:       The 'root' httpd server process will block, the machine will respond very slowly, and refuse all connections.
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    apache
>State:          closed
>Class:          sw-bug
>Submitter-Id:   apache
>Arrival-Date:   Mon Apr 21 22:30:01 1997
>Last-Modified:  Sun Jun 29 18:33:26 PDT 1997
>Originator:     joshr@stockmaster.com
>Organization:
>Release:        1.2b8
>Environment:
SunOS 4.1.4, few patches, built with gcc.  On a Sparc IPX running
about 50,000 invokations of a nph- cgi daily.
>Description:
I have observed the server to seriously misbehave in the following way.
I have not tried to reproduce it.  
Symptoms:
1)  top display showed the root httpd process at the top of the
	display, but not spinning cpus,  
4083 root      32    0  372K  276K run    14:51  1.96%  1.95% httpd

2) When I attached to the same process with gdb: Attaching to program `/usr/local/etc/httpd/httpd', process 4083sendmail
Reading symbols from /usr/lib/libc.so.1.9.1...done.
Reading symbols from /usr/lib/libdl.so.1.0...done.
0xf7737300 in fork ()
(gdb) where
#0  0xf7737300 in fork ()
#1  0x5680 in make_child ()
#2  0x60e4 in standalone_main ()
#3  0x6438 in main ()
(gdb) c
Continuing. (...no return...)

3) Then I traced the process:
race -p process
wait4 (0, 0xf7fffc94, 0x1, 0) = 15139
getpid () = 4083
gettimeofday (0xf7fffb08, 0) = 0
getpid () = 4083
gettimeofday (0xf7fffb08, 0) = 0
fork () = 15140
- SIGCHLD (20)
wait4 (0, 0xf7fffc94, 0x1, 0) = 15140
getpid () = 4083
gettimeofday (0xf7fffb08, 0) = 0
getpid () = 4083
gettimeofday (0xf7fffb08, 0) = 0
fork () = 15141
- SIGCHLD (20)
wait4 (0, 0xf7fffc94, 0x1, 0) = 15141
>How-To-Repeat:
I have no idea. This only happens very occasionally 
(I catch it occurring, say, once a month)
>Fix:

>Audit-Trail:
State-Changed-From-To: open-analyzed
State-Changed-By: dgaudet
State-Changed-When: Sun Apr 27 00:27:37 PDT 1997
State-Changed-Why:
There were related changes in the code for 1.2b9, if you could
give it a try when it comes out in a few days that'd be great.

Thanks
Dean
State-Changed-From-To: analyzed-closed
State-Changed-By: dgaudet
State-Changed-When: Sun Jun 29 18:33:26 PDT 1997
State-Changed-Why:
There were many race condition fixes around 1.2b10.  You should
try 1.2 release.  We're also recommending "KeepAlive off" in httpd.conf
for SunOS4 machines.

Dean
>Unformatted: