Re: Fedora Core 5 Issues

[alan <alan clueserver org>]

On Fri, 18 Aug 2006, Chong Yu Meng wrote:

> On Thu, 2006-08-17 at 17:48 +0200, Roger wrote:
>
> <snip>
> > I did not even have the Graphical aspect of Fedora installed i
> > installed just the bare minimum and for some reason a lot of things
> > were installed that i did not specify instance the Bluetooth, it was
> > causing my server to crash and i removed that, i removed exim
> > basically all i did was remove... remove... until the thing was making
> > my hard drives die.
>
> Hi Roger,
>
> Sorry to hear about your bad experiences. I am also running a production
> server on FC5, but I haven't had the same kind of stability problems you
> are facing. Of course, I am only running a web server on it right now --
> no X, no mail, no SATA, old Celeron motherboard.
>
> I find that it is always a good idea, after an install of any Linux (not
> just Fedora), to prune away all the packages that you will not need. Not
> just for stability, but also so that there are fewer vulnerabilities.
> Also, when setting up for certain applications, such as Oracle, you will
> need to tune kernel parameters, and for Java, previously you needed to
> add a parameter so it would not segfault. Network settings also
> frequently need to be adjusted. I always tell the younger engineers that
> Linux is not like Windows or even Solaris/AIX/HPUX, which seem to "just
> work" out of the box. Linux (any kind) for the server takes 2 days to a
> week to tune it just right -- even RHEL requires some pruning and
> tuning.

I always found Solaris or AIX to take longer to configure to a point where 
they are useful.  At least Linux comes with tools where the versions are 
from this century.  It usually takes 2+ weeks to get things built to what 
I consider a "usable" system.  (The prebuilt binaries that are available 
for those platforms tend to be either out of date, install in bizzare 
locations, or both.)

As for Windows, I don't find it useful at all.

I have not seen the types of stability issues this person is reporting.  I 
would suggest that they run memtest86 (www.memtest86.com) first and see if 
there is a problem there.  (Bad memory has been the cause of almost all of 
the stability issues I have seen.)

Turning up the logging on syslog to capture all kernel messages would be a 
good idea here.  Might give some clue as to what is failing and why.

--
"I want to live just long enough to see them cut off Darl's head and
 stick it on a pike as a reminder to the next ten generations that some
 things come at too high a price. I would look up into his beady eyes and
 wave, like this... (*wave*!). Can your associates arrange that for me,
 Mr. McBride?"
                      - Vir "Flounder" Kotto, Sr. VP, IBM Empire.