RE: Linux 2.2.12pre

Paul Jakma (Paul.Jakma@Digital.com)
Wed, 18 Aug 1999 13:57:24 +0100


> > Nobody uses the current knfsd in a production environment I think.
> >

even so, the knfsd tools shipped with RH6 is not comaptible with the more
recent knfsd utils.

> > However the difference is that once I convert a RAID partition
> > for use with an 2.2.12 kernel, I cannot boot earlier
> kernels anymore,
> > right? Or is that a misconception? That is _bad_ and worse it's
> > all undocumented
>

once you've booted a latest RAID kernel with old-raid drives you'll know all
about it though, and the conversion thing is pretty explicit.

> Perhaps the answer is that the next 2.2.x should skip to 2.4 and
> 2.3 should skip to 2.5? It seems clear that moving up the 2.2.x
> series does involve significant changes.
>

biggest problem imo is the reluctance to patch 2.2 up. before you took it
for granted that every now and then you'd have to update tools along with
the kernel. 2.2 has reversed the trend, and we've got to the situation where
the kernel stuff is way out of date compared to the maintainers patches. So
when the update finally happens it ends up being a big one, and people get
stung.

I've currently got various nfs, apm and tulip patches to apply when i build
a kernel. Thankfully with 2.2.12 i don't have to apply raid anymore. I also
had alpha patches to apply before 2.2.11. (but i still had to back out one
change...).

I also use devfs (but let's please not get into that), and would really love
to be able to use LVM on top of RAID, but that needs further patching.. :(.
All in all i spend ~30min applying patches and merging rejects when i build
a kernel. And most are so long standing that in a lot of cases i know the
rejects off by heart.

So why not be more adventurous with synching the kernel up. Those who want
to be on the edge can update their tools as neccessary, those who want tool
stability can track their distributions update instead. That'd be a lot
better than the current mess of endless patches to patches to ...

regards,

paul jakma.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/