Re: random system crashes

www2.incredimail.com is running on an IBM x330, 512MB ram, 2*PIII1000, 2
scsi 18GB disks in mirror.

serving tens of millions of hits per day in small files ( don't know exactly
how many, since turned off logging of tux), with backend apache and about 20
rules in iptables. running steady for months (using the last tux patch on 2
4.18 kernel from from kernel.org ). the only problem from time to time,
running tux stop after several months of uptime crashed my machine (happened
twice so far, but i can live with that). Outgoing bandwidth approaching
30-40Mbps, 4000-5000 concurrent connections, using only 10-15% cpu at daily
peaks. 

www5.incredimail.com is serving IncrediMail setup, 5 MB files, same hardware
 same stability, same outgoing bandwidth, up to 1000 concurrent clients at
peaks.
 
-------Original Message-------
 
From: [email protected]
Date: éåí ùðé 14 àå÷èåáø 2002 19:25:36
To: [email protected]
Subject: Re: random system crashes
 
Sounds to me as the old bug I saw one year ago.
Me, and as I know, some other people had the same experience (I dont know if
it was able to ping the machine because I had a rule in iptable to not
respond at ping)...
I thought someone had found what caused this, and in the next week I am
going to try tux on a productionserver again, but this sounds too bad...
 
The only thing in common with you is that my machine was a single-processor,
very high load, IDE-discs, and 1Gb ram. About 100 requests/second...
Normally it would run for 1 week, and then it totally crashed...
The bad behavoir stopped when I abandoned tux, running only at Apache (with
terribly bad performance as a result, but... )
 
Is here anyone out here who is running tyx successfully? A tux which is
using a backend-apache, and is server LOADS of small files every minute?
On what hardware?
 
//Andreas
----- Original Message -----
> From: "Marco Rothley" <[email protected]>
> To: <[email protected]>
> Sent: Monday, October 14, 2002 3:38 PM
> Subject: random system crashes
>
>
> > Hello all!
> >
> > I have a big problem with one of my webservers which is running TUX. I
> don't
> > know if TUX is the reason for that but the system crashes from time to
> time
> > and there is nothing special with it except the TUX.
> > The crashes are a little uncommon: The system reacts on nothing but
ping.
> Even
> > the local console isn't responding but the ping-timing is low as under
> usual
> > operation. There is nothing in the logfiles that points to a problem.
They
> > just stop and the next messages I see are the bootup messages of the
> kernel
> > (after reboot). One time I just had top running on the machine as it
> crashed
> > and I saw nothing special: Enough free memory, no high load and over 50%
> idle
> > time on the two CPUs, etc.. And from one moment to the other it was
> frozen.
> > I just thought someone could have similar problems and have a solution.
> The
> > hardware was completely replaced by the hoster but the problem persists.
> Just
> > the HD stayed the same but the RAID-Controller reported no failures. So
I
> > think it's not a hardware problem.
> >
> > The system is a dual P3 with 1.266 GHz, 1GB ECC-SDRAM from IBM (x330)
with
> a
> > SCSI RAID-Controller and 2x36GB IBM SCSI disks.
> >
> > The kernel Version is 2.4.18 with TUX 2.2.7. I don't know if there's any
> other
> > patch in the kernel. I have Apache as "backend" server.
> >
> > Thanks in advance!
> >
> > Marco.
> >
> >
> >
> > _______________________________________________
> > tux-list mailing list
> > [email protected]
> > https://listman.redhat.com/mailman/listinfo/tux-list
> >
>
 
 
 
_______________________________________________
tux-list mailing list
[email protected]
https://listman.redhat.com/mailman/listinfo/tux-list
 
.