On Mon, 11 Sep 2000, Jeff Garzik wrote:
> I hear that the new NFS patch is "better and more stable" etc. but no
> details.
hard to give details as i havn't used unpatched linux 2.2 nfs in a
very long time. best evidence from me is anecdotal: linux 2.4 / 2.2
nfs patches works perfectly for me (linux server, linux/irix clients,
V2 and V3), and Trond et al are very responsive if there's problem.
(eg silly delete handling bit me a couple of weeks ago, and Trond
redid it and had a patch for me within a week or so of diagnosing the
problem.).
What I remember from using unpatched linux nfs is that it was /awful/
compared to the other unixen i used, most notably in the stability
dept. As i remember it a standard linux NFS server never had much of
an uptime because of NFS. also dreadful performance to non-linux
clients.
> It seems to me that if you want an NFS problem fixed in 2.2.x,
> address that a single problem with a reproducible test and a small,
> focused kernel patch sent to Alan.
>
there is a patch (well a couple -> server side and client side). The
problem is it is never integrated, so now after more than a year of
development in NFS the difference between linux nfs and /current/ NFS
is bound to be huge.
> Whatever Alan's reasons for not including "the 2.2.x NFS patch", I
> serious doubt among those reasons is "keep NFS dreadful and unstable."
> ;-)
unless alan is being paid off by the commercial *nix vendors looking
to sell NFS server - no of course not. :)
> Maybe it breaks backwards compatibility...
not with current nfs-utils. And even if it did: show me anyone who
uses standard NFS for anything half-serious that isn't using the
patches. (either they're patching themselves or their vendor has
included the patch).
> so then, someone should pick up the ball and break up the NFS
> patch into acceptable, useful, tested chunks. Avoid the stuff
> that breaks backwards compatibility,
backward compat with what? we're now in the situation that the biggest
problem is that so vendors are using different revisions of the nfs
problems. none that i know of ship stock linux nfs.
even worse: we also have the situation that sometimes people submit
small patches against stock linux NFS to fix small problems that are
either fixed in the nfs patch or simply were never present in the nfs
patches.
> but submit the other fixes to
> Alan, describing in detail each problem fixed by each patch.
>
that'd be worth discussing on the nfs list... good idea.
> Jeff
--paulj
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Fri Sep 15 2000 - 21:00:17 EST