Re: Key features that should go into 2.4 ... (fwd)

Pedro Manuel (pedro@formtech.se)
Tue, 10 Aug 1999 21:34:48 +0200 (MDT)


I dont share your pessimistic view of nfs development in linux. I do
agree that it is important for a bigger linux acceptance in most
companies, but the current state is not bad, on the contrary. You have
working nfs v2, both as a client and a server. Nfs v3 is also available,
both as a client and a server, thanks to Trond Myklebust hard work, even
though i wouldnt put my hands in the fire for the server part. You talk
about nfs-interoperability with AIX, Irix and Sun. It exists, also. I have
Linux working as a nfs server for more than a dozen Sgi & Ibm
workstations, with heavy locking usage, without many complaints (not
perfect, but then, what is). Now, you ask me, what is the secret? I will
tell you - i have a coleague that once said to me - "its all about sex,
man, all about sex". Now i will reformulate it, and maybe even ask for a
patent on it - "its all about patches, man, all about patches". You need
the right ones, and fast. That gets us to the main problem here. Why arent
they in the stable kernel already? Well, i ask myself the same question
regarding nfs v2. Its something long due. I am waiting for 2.2.11 kernel
with hopes it will be included, even though i have yet to see the changes
list (i use 2.2.7, btw). Nfs v3 is a different beast. I think it needs to
be coded in a different way in order to be allowed in the kernel tree. It
will be done, i am sure, but it will take time.

Tell me what you need, where are you having problems. I will try to help
you.

Regards,

Pedro Rodrigues

>
> Hi all.
>
> First, let me say that I'm impressed with all the good work that you
> are doing. However, there are some parts where Linux is not
> "at the same level" as the rest of the kernel (and it's competitors).
>
> Althrough it's history Linux has been known for it's poor NFS support.
> Many of us has had hopes that this would be fixed in 2.2, but no. Now
> our hopes are set for 2.4, but the recent talk about a feature-freeze
> and hurrying a new kernel out have started a rumour that nfs in 2.4
> will remain unchanged.
>
> Today, Linux is excellent for anything that doesn't require nfs. That
> is www-servers, news-servers, samba-servers and all sorts of
> standalone-machines. You can use nfs, but only between Linux-machines
> if you don't want to trigger obscure bugs or have bad performance.
>
> Lots of environments rely on nfs for file-sharing, and today you often
> have to fight when you're going to use Linux as an nfs-client
> (especially with 2.2 which seems to have triggered various bugs in
> nfs-servers). You can almost forget to use Linux as an nfs-server for
> non-Linux machines though.
>
> I, and many many others, believe that if Linux are going to get
> accepted (and used) in such environments then it has to have a working
> nfs-support, both client and server. If the feature-freeze means no
> working nfs3 I hardly recommend you all to reconsider this. A (IMHO)
> working solution would be to do the feature-freeze, but push the
> development of the nfs-part into working state and into the kernel.
>
> It would definately be a win for Linux, since it's one of the weak
> points in the Linux-kernel (the weakest IMO). I could probably come up
> with many more reasons, but the best I can think of is that it would
> lead to far less frustrated sysadmins :-)
>
> If there is a need to test nfs-interoperability with AIX, Irix or
> Solaris feel free to drop me an email. Linux can talk to them, but
> Linux seems to trigger various bugs in AIX (and reportedly Irix, but
> we're not using any Irix nfs-server at the moment).
>
>
> If you are replying to this mail, please CC me since I'm not a
> subscriber of linux-kernel.
>
>
> /Nikke - SysAdmin at the Academic Computer Club, Umeå University, Sweden
> --
> -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
> Niklas Edmundsson, Admin @ {acc,hpc2n,ing}.umu.se | nikke@ing.umu.se
> ---------------------------------------------------------------------------
> Bluff means never having to sway your story.
> =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
>
>
> -
> 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/
>

-
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/