From owner-freebsd-stable Fri Oct 3 07:50:54 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id HAA18829 for stable-outgoing; Fri, 3 Oct 1997 07:50:54 -0700 (PDT) Received: from baklava.alt.net (root@baklava.alt.net [207.14.113.9]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id HAA18824 for ; Fri, 3 Oct 1997 07:50:51 -0700 (PDT) Received: from baklava.alt.net (ccaputo@baklava.alt.net [207.14.113.9]) by baklava.alt.net (8.8.5/8.7.3) with SMTP id HAA06951; Fri, 3 Oct 1997 07:50:34 -0700 (PDT) Date: Fri, 3 Oct 1997 07:50:34 -0700 (PDT) From: Chris Caputo To: dk+@ua.net cc: freebsd-stable@freebsd.org Subject: Re: Another NFS bogon in 2.2-stable? In-Reply-To: <199710030717.AAA05220@dog.farm.org> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-stable@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi. We are running "NetApp Release 4.1cD3: Wed Jul 23 20:07:56 PDT 1997". Are you heavily using the NFS server? We had several hundred processes accessing the NFS filesystems when the problem started happening. With light load there wasn't a problem. Our fix was to go back to 2.1-stable and fix some bugs in that NFS implementation. The same bug fixes may help 2.2-stable but we don't have the luxury of being able to test that right now. Chris On Fri, 3 Oct 1997, Dmitry Kohmanyuk wrote: > In article you wrote: > > We are definitely seeing problems with nfsv2 and either tcp or udp > > connections. We have tried the -r=1024 fix for udp and that didn't help. > > > We use the 2.2-stable box as an NFS client to a NetApp NFS server. After > > a little while, the FreeBSD box gets into a state where any command that > > causes NFS activity hangs, including a simple "df". Once in this state, > > the command can not be killed, even with "kill -9" from root. > > Which version of Data ONTAP (NetApp OS) are you running?? > > I have 4.1d talking to 2.2-stable clients and have never had this problem > with nfsv2 (with nfsv3, you get problems, like you cannot build the kernel > on NFS /usr/src/sys because vnode_if.c is built with 4K block of zero bytes. > (If you remove this file, and type make again, it works.) > I have default mount options, and use NetApp for /usr/cvs , /home , and > other actively-used filesystems. > > > I am gonna start comparing the 2.2-stable sources with 3.0-current sources > > to see if there are any obvious fixes. If anyone has any other ideas, > > please let me know. We have an immediate desire to get beyond this. ;-) > > -- > old unix hackers don't die, they just turn into zombie processes -- net.someone >