Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 4 Feb 2005 00:07:23 +0100
From:      "Daniel Eriksson" <>
To:        "'Mohan Srinivasan'" <>, "'Kris Kennaway'" <>
Subject:   RE: Processes stuck in nfsreq
Message-ID:  <!~!>
In-Reply-To: <>

Next in thread | Previous in thread | Raw E-Mail | Index | Archive | Help
Mohan Srinivasan wrote:

> Also, after you force a core, can you also try a quick 
> workaround - someone
> else also reported NFS client hangs and said that things were fine
> after they set mpsafenet to 0. It would be good to see if there's a
> correlation there.

That would be me. Unfortunately I've been busy with other things so I
haven't had time to switch back to mpsafenet=1 and get you a dump. Since
switching to mpsafenet=0 I haven't had a single NFS lock-up, so there seems
to be a correlation.

I have observed another strange thing lately. It seems like I keep getting
file corruption when transferring large files (10-75MB each) over NFS with
net.isr.enable=1. This is on an SMP client and a UP server, both running
very recent 6-CURRENT kernels hooked up using a crossover cable (if_em on
client, if_vr on server). The failure rate seems to be around 1 in 150 files
or something like that, and the error shows up as a file that is a few
hundred bytes shorter than the original (always resulting in a filesize on
an 8kB boundary). I only switched off net.isr yesterday, so I still don't
know for sure if it has cured the problem, but I've moved well over 1000
files since then without any corruption issues.

Again, the amount of details I can provide is very limited. Sorry about

/Daniel Eriksson