Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 10 Dec 1997 11:13:04 -0600
From:      Karl Denninger  <karl@mcs.net>
To:        "John S. Dyson" <toor@dyson.iquest.net>
Cc:        toasty@home.dragondata.com, freebsd-current@freebsd.org
Subject:   Re: 971208 and the nfs rollback didn't fix it...
Message-ID:  <19971210111304.33701@mcs.net>
In-Reply-To: <199712101626.LAA00918@dyson.iquest.net>; from John S. Dyson on Wed, Dec 10, 1997 at 11:26:13AM -0500
References:  <19971210082048.20706@mcs.net> <199712101626.LAA00918@dyson.iquest.net>

next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, Dec 10, 1997 at 11:26:13AM -0500, John S. Dyson wrote:
> Karl Denninger said:
> > > 
> > > A reboot advises me that not all processes died, check ps axl, then it locks
> > > up.
> > > 
> > > 
> > > This system is a dual P/200, acting as an nfs client....
> > > 
> > > Both machines(this client, and a 2.2.1 server) get random 'nfsd send error
> > > 55's on them...
> > > 
> > > Is this news to anyone?
> > > 
> > > Kevin
> > 
> > Post a "dmesg" with the boot log in it.  What ethernet card are you using?
> > 
> I am not an NFS "expert", so you guys tell me exactly what the symptoms are,
> and I will TRY to help work out the problems.
> 
> -- 
> John
> dyson@freebsd.org
> jdyson@nc.com

The posted "netstat -m" looks like an mbuf leak, and I'm not seeing them 
here (with VERY heavy load and use).  An example machine here:

11:11AM  up 8 days, 22:30, 37 users, load averages: 0.24, 0.31, 0.25

284/1280 mbufs in use:
	155 mbufs allocated to data
	129 mbufs allocated to packet headers
65/824 mbuf clusters in use
1808 Kbytes allocated to network (9% in use)
0 requests for memory denied
0 requests for memory delayed
0 calls to protocol drain routines

That's why I asked for the "dmesg" output and card in use - its possible
that the problem here is with the ethernet driver losing mbufs somewhere,
and if that's the case then the NFS code is not implicated at all.

That I haven't seen this kind of problem is relatively good evidence of
this - especially under our load profile.  I *am* running 1.41 of nfs_bio.c 
in my kernels, because later versions blow up very, very badly, and I haven't
rebuilt the master distribution since late November here.

--
-- 
Karl Denninger (karl@MCS.Net)| MCSNet - Serving Chicagoland and Wisconsin
http://www.mcs.net/          | T1's from $600 monthly to FULL DS-3 Service
			     | NEW! K56Flex support on ALL modems
Voice: [+1 312 803-MCS1 x219]| EXCLUSIVE NEW FEATURE ON ALL PERSONAL ACCOUNTS
Fax:   [+1 312 803-4929]     | *SPAMBLOCK* Technology now included at no cost



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19971210111304.33701>