Re: random system crashes

----- Original Message -----
From: "Marco Rothley" <[email protected]>
To: <[email protected]>
Sent: Saturday, October 19, 2002 3:46 PM
Subject: Re: random system crashes


> > this probably explains why there are only about 500 servers reported
using
> > tux on  netcraft. too bad development is on hold . People, please try to
> > help us troubleshoot this and determine what is the cause of the
problem -
> > give us more real life tests and experiences, i already done my tests,
but
> > we need to confirm this with other systems ...
> Hello!
>
> I would like to test this all, but I need the logging data. I'm still
running
> the same setup now since nine days without further hangs.
> log_incomplete is turned of in my enviroment. Should I try it with
> log_incomplete turned on? Would that help?

no, i don't believe so. we need a person who couldn't make tux work stable,
and can turn of logging completely to see if it solves the crashes. did your
system crash? if it did, was there some corellation to the log rotation time
?

> The problem is that you need a production server to test it under such a
load
> and without syntethic benchmarks. Most people don't like to test the
> stability of software on a production system. So we need a reproducable
case
> where TUX hangs. The problem with that is that as far as I can see the
hangs
> only occures with SMP-systems (am I right? Is there an example with a
single
> CPU system?)

i am doubted that ppl with the kind of traffic we need will base their
server on a single cpu machine, and any half-decent server cones with 2
cpu's nowadays, but if i am wrong, i will happyly accept the data

> and I don't have one that isn't used as an production server. So
> I can't elaborate a test case...

we need a production system with a standby computer, so if the main crashes,
the backup takes over automatically, like i have. I already ran all possible
tests, i need data from other sources now .

> Marco Rothley