Adventures with Fedora in Biotech

[Bevan C. Bennett]

Hugh Caley wrote:

> We had problems with it in a server context, however.  We had set up a 
> couple of dual 2.8 mhz Xeon boxes with 4 gig of ram for running 
> bioinformatic perl scripts.  The results were disappointing; in many 
> cases the scripts would run at half the speed they would run on other 
> slightly slower (but still Xeon 2.x) servers which were running RedHat 8 
> or 9.  We tried  changing /etc/sysconfig/i18n to use C or en_US as the 
> default language, and unlike RedHat 9 it didn't seem to make much of a 
> difference.  Also, the servers were unstable and would lock up after a 
> few days of pounding perl.  We had to put RH 9 on them and now they 
> don't go down and run at full speed.  Weird.  Not really surprising, I 
> guess, with the first release of Fedora.
> FYI.

That is odd. I have a dozen 2.8GHz Xeon boxes (only 2G of RAM each) that 
I use as a compute farm and they've been completely happy and stable 
under Fedora Core 1 (up 16 days since last kernel update).

I did try to put a 2.6.1 kernel on one of them, but ran into a lot of 
trouble with that and eventually gave up and stuck with stock FC1.

I don't have any perl performance benchmarks, but the FC1 systems 
maginally faster at our number crunching (Java based) than they were before.

Just adding another data point.