Date: Mon, 22 May 2000 13:16:46 -0700 (PDT) From: Doug Barton <Doug@gorean.org> To: Matthew Dillon <dillon@apollo.backplane.com> Cc: freebsd-hackers@FreeBSD.ORG Subject: Re: NFS server problems on 3.4-S, any interest? Message-ID: <Pine.BSF.4.21.0005221246160.25742-100000@dt051n0b.san.rr.com> In-Reply-To: <200005221724.KAA61109@apollo.backplane.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, 22 May 2000, Matthew Dillon wrote:
> : I have two machines at home, one running 3.4-Stable which is my
> :firewall/natd/file server machine. The other is running 5.0-Current
> :(around 5/8 right now) which mounts a whole bunch of stuff via amd from
> :the 3.4 machine. Up until say... 2 months ago, this worked really well.
> :I never had any nfs performance problems. One of the things I mount for
> :example is /usr/src, and my make world time didn't suffer much, if at
> :all.
> :
> : However, at some point about 2 months ago (I wish I had a more definite
> :time period, but the problem has kind of snuck up on me) I have been
> :getting increasingly bad performance from the nfs mount. I get "nfs
> :...
> :Doug
>
> First, make sure the 5.x machine is up to date.
I'm going to upgrade it today, I'm checking out a new source tree
local to the workstation machine.
> Second, start looking for network problems, because if you are getting
> 'host not responding' errors it's probably due to packet loss somewhere.
> NFS is particularly sensitive to packet loss. This doesn't sound like
> an NFS problem on the face of it.
The errors are of the sort:
workstation /kernel: nfs server fileserver:/usr/ncvs: not responding
It's possible that it's a network problem I suppose, although I'm
using a crossover cable from an fxp0 in the workstation to an xl0 in the
fileserver. Both are running at 10Mb, full duplex. Netstat -id shows:
From the workstation:
Name Mtu Network Ipkts Ierrs Opkts Oerrs Coll Drop
fxp0 1500 <Link#1> 32102492 0 31653667 0 30900 0
From the fileserver:
Name Mtu Network Ipkts Ierrs Opkts Oerrs Coll Drop
xl0 1500 <Link> 32504173 28967 32900227 0 0 0
I did find it a little unusual that I was getting collisions on a
crossover cable, but when I looked at the mail archives related to that
problem I read that the intel cards are very aggressive packet pushers,
and that this isn't all that unusual. The ratio of good packets to
collisions seemed healthy enough to not warrant too much concern.
Hrrrmm... I just took a look at the settings for each card. I did
not specify full duplex in the fxp0 ifconfig line, since autoselect has
always worked before. I ran 'ping -f' simultaneously on both hosts, and
did experience a 4% packet loss on one host, and 1% on the other. Changing
to an explicit:
ifconfig fxp0 inet 10.0.0.2 netmask 255.255.255.0 media 10baseT/UTP
mediaopt full-duplex
gets me 0 packet loss on both sides. So you may be right Matt, it may be a
network error after all. I'll give it another test run and see what I can
see.
In case it's significant, the fileserver has been up for 11 days
(last make world) and the workstation has been up for two.
Thanks,
Doug
--
"Live free or die"
- State motto of my ancestral homeland, New Hampshire
Do YOU Yahoo!?
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.21.0005221246160.25742-100000>