Hi again!

Thanks for the responses.... 
I don't have a spare box with similar hardware..... the cluster is very 
heterogeneous.... we have some PII 400,450,500 (may be 6 or 7) and some 
celerons with sis, mach,etc video cards.... a duron and a athlon.... with 
some generic TNT2 video cards..... I know I could solve the config problems 
that could arise. But the other problem is that a machine is the nis 
server..... the other the dns server.... another is the printing server. I 
distribute the load so if something fails not the hole system goes down, 
and we could use all the machines to run jobs.
We use it to make computational simulations.... and the physicist are 
difficult people when they can't use the cluster.. ( I'm a physicist too :) ).
I already use the idea of put the .rpms in one directory and the rpm -Fhv the 
hole directory.... that's the way I use to upgrade till now. I share a 
directory between the machines (/usr/loca/) with nfs. 

May be I could find a spare machines to test an upgrade.... and somehow rsync 
with the rest... (I must use a couple of machines because I would need to 
test the response from this machine with the diskless machines that use this 
to boot).

Meanwhile I will try to compile glibc 2.2.4 and try to install it in another 
directory... (may be under /usr/local/ or /opt/lib that is shared between the 
machines). And test some precompiled binaries (may be extract some binaries 
from the rpm packages) that need the 2.2.4 library, changing the LDPATH 
environmet variable. 
I hope it didn't explode. 


Thanks!!!
ALF

Want to buy your Pack or Services from MandrakeSoft? 
Go to http://www.mandrakestore.com

Reply via email to