[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [cobalt-users] Have SUN stopped supporting XTR?



Gerald Waugh wrote:

> > 1. Figure out how to reliably flash the firmware so that we may use 2.4
> > kernels on the boxes.
> 
> I think Tim Hockin has this down!
>   ftp://ftp-eng.cobalt.com/pub/users/thockin/2.4/

I thought that this occasionally or even regularly created doorstops. I
distinctly remember several messages in the past where people ended up killing
their boxes after upgrading the firmware failed. If the process has been
improved I'm all for it.


> > 2. Get a working 2.4 kernel (I think Sun has these somewhere in the
> > depths of their ftp site along with the firmware upgrades). By working I
> > mean the whole thing, including the front panel.
> 
> Again ftp://ftp-eng.cobalt.com/pub/users/thockin/2.4/

Sounds like getting Tim on board would be nice. :) Too bad these aren't srpms.
Installing a kernel as an RPM in either a Debian or Gentoo box at the very least
would be a pain. Installing a generic 2.4.x kernel and then working on getting
the panel working in some fashion may be easier. I beleive that the only other
thing that gets patched in the kernel is the bwmgmt stuff and I thought that
there were other tools for this these days. I wonder though how many people use
the bandwidth management stuff, and if we wouldn't be better off documenting how
to do it by tweaking proftpd/apache config files (I'm pretty sure that given the
right modules both of them can throttle bandwidth).


> Yes, I don't think you can use their "proprietary stuff" that is why we
> are working on these candidates -  debian, slackware or gentoo port

The raqs are too underpowered to be good Gentoo canidates. It's a great OS, but
recompiling large packages on slow boxes can be tedious at best. I believe that
you can compile them once and then shove out the binaries to other hosts, but
that's likely experimental at best. I think that Debian would be the best
canidate. I suppose a well written web interface to the apt tools would be the
equivalent of a working BlueLinQ.


> I don't think we want to start a project that will duplicate the work already
> being done by many good people.

Other than Gentoo I've never been satisfied with the up-to-date-ness of a
distro. Debian is probably good enough, but the long time between releases got
annoying. I imagine that a semi-yearly 'apt-get dist-upgrade' whenever Debian
gets around to updating is good enough for most people. I'd rather be able to
run the latest version of Apache+PHP+Mod_ssl, Proftpd, MySQL, OpenSSH, etc. at
all times. I suppose for those of us interested in that we can just maintain
separate .debs of those packages with less fear of them being required by an
unrelated patch from Cobalt (ie. you're already running the latest and greatest
version of PHP, but now you're forced to downgrade in order to install a patch
to bind).

> Again, there are distros that are well supported, Sun/Cobalt being the
> exception. I like their hardware!

I got a stack of fans and drives that say otherwise... ;)

 
> Webmin is not that bad as a user GUI.

It must have improved then because it used to be pretty bad. I'll take another
look.


> Youre talking about one hell of a lot of work Travis.
> Your intentions are excellent! But I really think that getting the
> hardware to run a well supported distro is the way to go.

Probably. See my other message on the subject. As long as we can get a modern
distro running and still maintain things like the front panel, and the ease of
OS restores (two things I'm certain that given enough time that we could do) I'm
in. 

-- Travis