Re: Getting rid of old, obsolete kernels

• Previous message: Michael Heiming: "Re: Red Hat 9 won't log me in"
• In reply to: Michael Heiming: "Re: Getting rid of old, obsolete kernels"
• Next in thread: Nico Kadel-Garcia: "Re: Getting rid of old, obsolete kernels"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Date: Sun, 24 Jul 2005 12:43:43 -0400

"Michael Heiming" <michael+USENET@www.heiming.de> wrote in message
news:vhuar2-fu9.ln1@news.heiming.de...
> In comp.os.linux.setup Nico Kadel-Garcia <nkadel@comcast.net>:
>
>> "Michael Heiming" <michael+USENET@www.heiming.de> wrote in message
>> news:qgear2-bsp.ln1@news.heiming.de...

>> Ahh, Sun's are their own special class. But I've seen several BIOS's in
>
> Not at all, those are just 19" rack server like the others. Just
> pointed them out because of the excellent integration.

Their own special class refers to the quality under the hood, Sun's have
been very solid for quite some time. They actually properly test their
BIOS's and components. I like Sun's, I just can't afford them for the sorts
of deployments I've done. When you're installing 1000 machines, it's often
cheaper and in some ways more reliable to install inexpensive hardware and
hire some poor beggar to actually learn their ins and outs and do deep
support for them, in-house.

> Might happen, but why bother, open support call, let the vendor
> upgrade them, at least they installed the buggy bios in the first
> place and will take care of console settings.

Umm. I was the guy answering the tech support call, or the guy who had to
make the tech support call, and find that various vendors had changed
components without notifying us. Having to test motherboards with alpha or
beta BIOS updates to address such issues is.... pretty painful when the
systems are not quite what you originally ordered or what you originally
tested with before buying the machines.

>> Yes, it should. But if you believe that all motherboard vendors can be
>> relied on to conserve such features reliably across BIOS versions or
>> across
>> minor revisions of the motherboard which your server or desktop vendor
>> may
>> not even tell you has been altered, I think you're in for a harsh
>> surprise.
>
> No problems if you restrict vendors to the lowest possible. Sure
> a zoo might give you some headache. As this thread...

Hmm. You mean fewest number of vendors? I've seen such issues with even
large, reputable suppliers, and had it happen with one of the best known and
reputable venors in the business. Hardware changes, not the BIOS failures:
it led to a large amount of upset in a big contract, ameliorated by the fact
that my kernel updates to support lots of newer hardware was actually due to
be installed the next week. I have a serious rant or two about what our
kernel people were doing to the source trees, but it wasn't the first time
they'd failed to merge changes from one source tree or another back to the
original code line and seriously, seriously broken things.

• Previous message: Michael Heiming: "Re: Red Hat 9 won't log me in"
• In reply to: Michael Heiming: "Re: Getting rid of old, obsolete kernels"
• Next in thread: Nico Kadel-Garcia: "Re: Getting rid of old, obsolete kernels"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]