tux on 2.4.27 kernel and referrer checking

[William Lovaton]

Cool!

I guess that a server is stable (not only 2.6) just when it is doing
very specific tasks.  In my case it runs an enterprise web app with
Apache, PHP, Oracle (The DB is in another server).  Around 900
concurrent users.  This is 99.9% of the usage pattern.

Another little web app installed in that server needs to send mail so it
uses sendmail too and uses and alternate mail server to make the actual
delivery.

It has xinetd activated with telnet, sshd and ftp (for remote admin
purposes).  This is an intranet server BTW.

And from time to time I run X apps to show them in a remote display (my
workstation) to check some statistics and do some maintenance (gkrellm,
system-config-* tools, etc).

May be there are people running lots of things in a single hardware,
let's say: Apache/PHP, Tomcat, MySQL, PostgreSQL, Any mail server, FTP,
Samba, NFS, Squid, IRC, etc, etc.  And suppose that the usage pattern is
equally distributed between all the services.

Obviously, under such scenario is more posible to get caught in a system
crash.  But even so I think almost any linux kernel should be up to the
task.


-William


El jue, 28-10-2004 a las 10:33, joe escribió:
> I define stable as how a given kernel performs day in and day out in 
> actual use on a busy server. Ironically, I moved to the 2.5 test series 
> awhile back because our trusty redhat 8 firewall kept locking up at 
> random, and replacing the vendor supplied 2.4 kernel with a late 2.5 
> series test kernel actually stabilized it.
> 
> I have been deploying suse servers with 2.6 kernels in production and 
> found them to be stable - IOW the servers don't generally go down unless 
> it's for kernel upgrade or hardware maintenance. There was one nasty 
> problem with the hbdev virus scanner, which loads a kernel module, and 
> would do nasty things to capabilities, prevent bind from starting, and 
> hang the system if you tried to shut it down - I chalk that up to bad 
> code, and use clamav instead. Otherwise the 2.6 systems I've deployed 
> seem to be rock solid.
> 
> I also found fedora core with 2.6 to be stable as well - I tried the 2.6 
> version of tux, and it did not perform all that well, so I have not 
> really done much with it since, as the general kernel improvements have 
> caused apache performance to come into the acceptable range.
> 
> Joe
> 
> 
> William Lovaton wrote:
> 
> >Very interesting discussion...  A question for all of you: How do you
> >define "stable"?  How do you measure it?  Have you seen crashes with 2.6
> >kernels?  Are they reproducible?
> >
> >I'm using Fedora Core 2 (with official updates) in a high loaded, high
> >traffic production server and it is very, very stable.  Right now it has
> >25 days of uptime.  It could be more by now, but some reboots have
> >prevented it.
> >
> >The only problem I have is TUX (not using it right now) and that's why
> >I'm subscribed to this list.  Anyway TUX is not present in the official
> >kernel anymore.
> >