Re: crashes / khttpd

Actually, I am running tux with logging disabled at all - otherwise it would
fill up 10 GB disk in 1 day .

People who responded earlier about successfully running tux, do you have
logging enabled ?

and please, please, do not state that the version of tux you are running is
2.2.7 or any other such number. This is the version of the userspace
utilities ONLY. Tux version is in the /include/tux.u.h file of the kernel,
and the latest is 2.1.1, although i am not sure that it get's updated
whenever a new kernel patch is released, so the best way is just to state
that you are running the 
tux2-full-2.4.18-final-A3 patch.

Thank you.
 
-------Original Message-------
 
From: [email protected]
Date: éåí ùìéùé 15 àå÷èåáø 2002 17:42:34
To: [email protected]
Subject: crashes / khttpd
 
Hello again!
 
Thank you all for the huge amount of response! I really didn't expect this 
because the list is usually quite quiet. It seams that there are many 
experienced people on that list how seldom have questions, but when there is

a question there are a many answers in a very short time, too!
 
I just wanted to add some information that might be helpful:
I didn't install that maschine by myself so I can't tell you the exact 
configuration. It is a "managed" solution from NTT/Verio here in germany and

they have there own distribution based on Red Hat. As far as I heard from 
people from Verio they used a vanilla 2.4.18 kernel and just their own 
configuration. The only patch they applied is the TUX-patch (2.2.7).
One thing that might be interessting is that the server never crashed under 
"normal" operation. Everytime it crashed someone was logged in with ssh and 
edited some text files or executed other commands. But nothing special. The 
system crashed 6 times in one week with the old hardware and after a
hardware 
exchange (another x330 but with two 1266 MHz CPUs instead of the two 1133
MHz 
and the same HDs) it just crashed one time in one and a half week. But the 
symptoms were identical to those of the crashes on the old maschine so I 
think the reason should be the same, too.
The system serves about 3 million hits a day (up to ~10 mil. on good days)
and 
is the main webserver of the German online shop "www.dress-for-less.de". The

shop uses apache 1.3.27 with mod_perl and I need a small (in memory 
footprint) and fast webserver for the many product pictures and other static

stuff like css-files and JavaScript etc. The optimal solution is the one
with 
TUX and all my tests with other systems were positiv so I decided to use it 
on a production server.
The configuration of TUX is standard (client port 8080 and TUX listening on 
port 80) except the logging. I need the referers in the log but don't want 
the redirects logged because I have them in my apache-logs anyway. Could the

crashes be related with the TUX-logger? So I have:
echo 1 > /proc/sys/net/tux/logging
echo $LOGFILE > /proc/sys/net/tux/logfile
#where $LOGILE is /usr/local/apache/logs/tux
echo 0 > /proc/sys/net/tux/redirect_logging
echo 1 > /proc/sys/net/tux/referer_logging
Maybe the referer logging is buggy (haven't looked at the source yet)?
If I try lsmod I get:
lsmod: QM_MODULES: Function not implemented
I don't know much about the kernel but it seems that there is something 
missing?!
 
If there will be more system crashes in the near future I will have to 
consider using thttpd. I read the khttpd mailinglist too und it would be 
enough for my purposes but the project seems to be nearly dead. The last
post 
on the mailinglist is from 19.8.2002 and than again at 22.5.02. There has 
been discussion to put khttpd out of the standard kernel on that list and 
there are massive problems with khttpd and SMP-kernels so I don't think that

khttpd would be better than TUX.
 
Thanks!
 
Marco.
 
 
 
_______________________________________________
tux-list mailing list
[email protected]
https://listman.redhat.com/mailman/listinfo/tux-list
 
.