Re: WWW Form Bug Report: "hangs shortly after startup, unable to kill" on UnixWare

[Chuck Murcko <ch...@telebase.com>]

Hi Jamie. Did you build Apache with these flags:

AUX_CFLAGS= -DSVR4 -DNEED_LINGER
AUX_LIBS= -lsocket -lnsl -lcrypt

???

jamie@nizkor.almanac.bc.ca liltingly intones:
> 
> Submitter: jamie@nizkor.almanac.bc.ca
> Operating system: UnixWare, version: 2.0
> Version of Apache Used: 1.1.1
> Extra Modules used: status
> URL exhibiting problem: 
> 
> Symptoms:
> --
> A few minutes after starting up httpd for the first time, it hung. It
successfully delivered a number of web pages for about five minutes,
then died.  I did a "kill -HUP 1234" and there was no apparent change,
so I did a "kill -TERM 1234".  That produced a "shutting down" message
in the error log, and the main httpd process went away -- but its nine
child processes were left hanging around.

su'ing to root and kill'ing the child processes does nothing, even
"kill -9".

Since the child processes think they own port 80, I am incapable of
starting up another httpd.  I'm working remotely so I can't restart
the system.  In short, I'm screwed.  Argh.  Very frustrating.

Based on the logs, I'm guessing that it's the status module that's
screwing up somehow, probably because I'm on UnixWare 2.0 which someone
said wasn't well-tested, possibly having something to do with the
"refresh" option.  That's my guess because a "status/?refresh" is the
last entry in the log.

Here are the last!
!
>   few entries in the access log:

hd04-146.compuserve.com - - [10/Jul/1996:12:34:28 -0700] "GET /rue/RUE3-ZuendelBio.html HTTP/1.0" 404 -
clmx12.dial.voyager.net - - [10/Jul/1996:12:34:59 -0700] "GET /status/ HTTP/1.0" 200 2022
clmx12.dial.voyager.net - - [10/Jul/1996:12:36:38 -0700] "GET /status/?refresh=20 HTTP/1.0" 200 2036
clmx12.dial.voyager.net - - [10/Jul/1996:12:37:06 -0700] "GET /status/?refresh=20 HTTP/1.0" 200 2034

(voyager.net is me.)

Another possibility is that there was some kind of infinite loop because
I'd set up error 404 to point to a file that didn't exist.  That's less
likely because it happened twice without apparent problems.  Note the
timestamp on the last error is fully two minutes before the last
successful GET command, above.  Here are the last few entries in the
error log, ending with the SIGTERM I sent it to try to shut it down:

[Wed Jul 10 12:26:25 1996] access to /web/orgs failed for homer.infoseek.com, reason: File does not exist
[Wed Jul 10!
!
>   12:26:25 1996] access to /web/errors failed for homer.infoseek.com,re
> 
> ason: File does not exist
[Wed Jul 10 12:26:37 1996] send lost connection to client pm042-02.dialip.mich.net
[Wed Jul 10 12:34:27 1996] access to /web/rue/RUE3-ZuendelBio.html failed for hd04-146.compuserve.com, reason: File does not exist
[Wed Jul 10 12:34:28 1996] access to /web/errors/error-404.html failed for hd04-146.compuserve.com, reason: File does not exist
[Wed Jul 10 12:55:05 1996] httpd: caught SIGTERM, shutting down

Any ideas would be welcome.

My Configure and *.conf files are available at:

ftp://ftp.almanac.bc.ca/pub/miscellany/apache/Configuration.almanac
ftp://ftp.almanac.bc.ca/pub/miscellany/apache/access.conf
ftp://ftp.almanac.bc.ca/pub/miscellany/apache/httpd.conf
ftp://ftp.almanac.bc.ca/pub/miscellany/apache/srm.conf
> --
> 
> Backtrace:
> --
> 
> --
> 

chuck
Chuck Murcko	N2K Inc.	Wayne PA	chuck@telebase.com
And now, on a lighter note:
The grand leap of the whale up the Fall of Niagara is esteemed, by all
who have seen it, as one of the finest spectacles in nature.
		-- Benjamin Franklin.