<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body text="#000000" bgcolor="#ffffff">
On 07/23/2010 05:15 PM, Gavin Hurlbut wrote:
<blockquote
cite="mid:AANLkTimh+MEQZ3E_mnHr-V6WssQspo50DxKz2gh4t-vX@mail.gmail.com"
type="cite">
<div class="gmail_quote">
<blockquote class="gmail_quote"
style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">intr:
deprecated<br>
</blockquote>
<br>
<div><br>
</div>
<div>Not deprecated in my man page. </div>
<div><br>
</div>
</div>
</blockquote>
<br>
"The intr / nointr mount option is deprecated after kernel 2.6.25.
Only SIGKILL can interrupt a pending NFS operation on these kernels,
and if specified, this mount option is ignored to provide backwards
compatibility with older kernels."<br>
<br>
Maybe you're using an older distro, like Debian Stable or something.
Or something non-Linux, like Solaris or BSD.<br>
<br>
<blockquote
cite="mid:AANLkTimh+MEQZ3E_mnHr-V6WssQspo50DxKz2gh4t-vX@mail.gmail.com"
type="cite">
<div class="gmail_quote">
<div> </div>
<blockquote class="gmail_quote"
style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">rsize/wsize:
The "TRANSPORT METHODS" section of the man page says you should specify
rsize and wsize when using UDP, not when using TCP. And all modern NFS
implementations use TCP.<br>
</blockquote>
<div> </div>
<blockquote class="gmail_quote"
style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">nfsvers:
autonegotiated to 3 if both sides support it, otherwise 2. So this
seems unnecessary also.<br>
</blockquote>
<div><br>
</div>
<div>If your NFS server doesn't require the options, great.
However... If it does NOT, you should do it...</div>
<div><br>
</div>
<div>nfsvers=3 is a minimum. If both sides do not support 3, you
want the mount to FAIL. NFSv2 does not support files > 2GB in size,
making it absolutely useless for mythtv recordings.</div>
<div><br>
</div>
</div>
</blockquote>
<br>
Okay, sounds good to me. :)<br>
<br>
<blockquote
cite="mid:AANLkTimh+MEQZ3E_mnHr-V6WssQspo50DxKz2gh4t-vX@mail.gmail.com"
type="cite">
<div class="gmail_quote">
<div>As for the rsize/wsize... I don't care what your man page
indicates, those are the parameters to tweak for performance in NFS.
If your setup defaults to good settings, great. I would not presume
that "all modern...." means squat. I have seen "modern" NFS
implementations that use UDP, and that require the options stated. </div>
<div><br>
</div>
<div>The *default* rsize/wsize for many NFS servers is 4k. This
sucks really badly for performance, especially on GigE.</div>
</div>
</blockquote>
<br>
My NFS server *does* give good defaults. Here is what I get mounting
with just the "rw" option, and exported with "rw,async":<br>
<br>
Flags: rw, relatime, vers=3, rsize=262144, wsize=262144, namlen=255,
hard, proto=tcp, timeo=600, retrans=2, sec=sys, mountaddr=192.168.1.2,
mountvers=3, mountport=37760, mountproto=udp, addr=192.168.1.2<br>
<br>
I don't have as much experience with NFS as you obviously do. I'm only
going by my man page and the results I see on my system. I believe all
modern Linux distros will see similar results, since they all use the
same NFS implementation, namely the one built into the Linux kernel.
Obviously Solaris, BSD, old userspace Linux NFS, etc. may be
implemented differently.<br>
<br>
It's Friday, and I've been in front of a computer long enough for
today. Have a good weekend!<br>
<br>
-- Kevin<br>
<br>
</body>
</html>