> Anyway I cannot duplicate their lockup. My ethernet spent the entire
> night jammed serving pages off apache 1.1.3 with no keepalives doing
> as many requests in parallel as will fit.
The InfoWorld machine was getting hit at 100mbits on three DE500 cards,
simultaneously. They wanted to use four, but they couldn't get a kernel
to recognize all four (though we had it working w/ four before the machine
left).
One of the big problems with this review was the SMP problems. In hindsight,
it was a *big* mistake for us to send them an SMP machine. That caused them
to have to reinstall the machine, and configure everything themselves. When
they did this, they lost the fourth ethernet card.
The people who did this review were *not* Linux hackers. I'm actually pretty
pleased with the review, as it did focus on the working machine. They could
have complained about the numerous hardware and software problems they had,
but they really did focus on the positives.
I do apologize for this grey-eye for Linux (calling it a black eye seems a
bit harsh); we at Red Hat should have been more agressive in working with
InfoWorld on this. I'd be happy to address any specific questions on the
machine or testing methodology, though I may not know all of the answers.
Please ask them in private email however, as I don't often read this list.
Erik
-------------------------------------------------------------------------------
| "Psychopaths kill for no reason: I kill for money." -- Grosse Pointe Blank |
| |
| Erik Troan = ewt@redhat.com = ewt@sunsite.unc.edu |