From: "Andrew Theurer" <habanero@us.ibm.com>
To: "Hirokazu Takahashi" <taka@valinux.co.jp>
Cc: <neilb@cse.unsw.edu.au>, <davem@redhat.com>,
<linux-kernel@vger.kernel.org>, <nfs@lists.sourceforge.net>
Subject: Re: Re: [PATCH] zerocopy NFS for 2.5.36
Date: Thu, 17 Oct 2002 08:16:43 -0500 [thread overview]
Message-ID: <003301c275e1$0bf76810$2a060e09@beavis> (raw)
In-Reply-To: 20021017.113126.102592502.taka@valinux.co.jp
Subject: Re: [NFS] Re: [PATCH] zerocopy NFS for 2.5.36
> Hello,
>
> Thanks for testing my patches.
>
> > I am still seeing some sort of problem on an 8 way (hyperthreaded 8
> > logical/4 physical) on UDP with these patches. I cannot get more than 2
> > NFSd threads in a run state at one time. TCP usually has 8 or more.
The
> > test involves 40 100Mbit clients reading a 200 MB file on one server (4
> > acenic adapters) in cache. I am fighting some other issues at the
moment
> > (acpi wierdness), but so far before the patches, 82 MB/sec for NFSv2,UDP
and
> > 138 MB/sec for NFSv2,TCP. With the patches, 115 MB/sec for NFSv2,UDP
and
> > 181 MB/sec for NFSv2,TCP. One CPU is maxed due to acpi int storm, so I
> > think the results will get better. I'm not sure what other lock or
> > contention point this is hitting on UDP. If there is anything I can do
to
> > help, please let me know, thanks.
>
> I guess some UDP packets might be lost. It may happen easily as UDP
protocol
> doesn't support flow control.
> Can you check how many errors has happened?
> You can see them in /proc/net/snmp of the server and the clients.
server: Udp: InDatagrams NoPorts InErrors OutDatagrams
Udp: 1000665 41 0 1000666
clients: Udp: InDatagrams NoPorts InErrors OutDatagrams
Udp: 200403 0 0 200406
(all clients the same)
> And how many threads did you start on your machine?
> Buffer size of a UDP socket depends on number of kNFS threads.
> Large number of threads might help you.
128 threads. client rsize=8196. Server and client MTU is 1500.
Andrew Theurer
-------------------------------------------------------
This sf.net email is sponsored by: viaVerio will pay you up to
$1,000 for every account that you consolidate with us.
http://ad.doubleclick.net/clk;4749864;7604308;v?
http://www.viaverio.com/consolidator/osdn.cfm
_______________________________________________
NFS maillist - NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs
WARNING: multiple messages have this Message-ID (diff)
From: "Andrew Theurer" <habanero@us.ibm.com>
To: "Hirokazu Takahashi" <taka@valinux.co.jp>
Cc: <neilb@cse.unsw.edu.au>, <davem@redhat.com>,
<linux-kernel@vger.kernel.org>, <nfs@lists.sourceforge.net>
Subject: Re: [NFS] Re: [PATCH] zerocopy NFS for 2.5.36
Date: Thu, 17 Oct 2002 08:16:43 -0500 [thread overview]
Message-ID: <003301c275e1$0bf76810$2a060e09@beavis> (raw)
In-Reply-To: 20021017.113126.102592502.taka@valinux.co.jp
Subject: Re: [NFS] Re: [PATCH] zerocopy NFS for 2.5.36
> Hello,
>
> Thanks for testing my patches.
>
> > I am still seeing some sort of problem on an 8 way (hyperthreaded 8
> > logical/4 physical) on UDP with these patches. I cannot get more than 2
> > NFSd threads in a run state at one time. TCP usually has 8 or more.
The
> > test involves 40 100Mbit clients reading a 200 MB file on one server (4
> > acenic adapters) in cache. I am fighting some other issues at the
moment
> > (acpi wierdness), but so far before the patches, 82 MB/sec for NFSv2,UDP
and
> > 138 MB/sec for NFSv2,TCP. With the patches, 115 MB/sec for NFSv2,UDP
and
> > 181 MB/sec for NFSv2,TCP. One CPU is maxed due to acpi int storm, so I
> > think the results will get better. I'm not sure what other lock or
> > contention point this is hitting on UDP. If there is anything I can do
to
> > help, please let me know, thanks.
>
> I guess some UDP packets might be lost. It may happen easily as UDP
protocol
> doesn't support flow control.
> Can you check how many errors has happened?
> You can see them in /proc/net/snmp of the server and the clients.
server: Udp: InDatagrams NoPorts InErrors OutDatagrams
Udp: 1000665 41 0 1000666
clients: Udp: InDatagrams NoPorts InErrors OutDatagrams
Udp: 200403 0 0 200406
(all clients the same)
> And how many threads did you start on your machine?
> Buffer size of a UDP socket depends on number of kNFS threads.
> Large number of threads might help you.
128 threads. client rsize=8196. Server and client MTU is 1500.
Andrew Theurer
next prev parent reply other threads:[~2002-10-17 13:22 UTC|newest]
Thread overview: 90+ messages / expand[flat|nested] mbox.gz Atom feed top
2002-09-18 8:14 [PATCH] zerocopy NFS for 2.5.36 Hirokazu Takahashi
2002-09-18 23:00 ` David S. Miller
2002-09-18 23:54 ` Alan Cox
2002-09-18 23:54 ` Alan Cox
2002-09-19 0:16 ` Andrew Morton
2002-09-19 2:13 ` Aaron Lehmann
2002-09-19 3:30 ` Andrew Morton
2002-09-19 3:30 ` Andrew Morton
2002-09-19 10:42 ` Alan Cox
2002-09-19 10:42 ` Alan Cox
2002-09-19 13:15 ` [NFS] " Hirokazu Takahashi
2002-09-19 20:42 ` Andrew Morton
2002-09-19 21:12 ` David S. Miller
2002-09-19 21:12 ` [NFS] " David S. Miller
2002-09-21 11:56 ` Pavel Machek
2002-09-21 11:56 ` Pavel Machek
2002-10-14 5:50 ` Neil Brown
2002-10-14 6:15 ` David S. Miller
2002-10-14 10:45 ` kuznet
2002-10-14 10:48 ` David S. Miller
2002-10-14 12:01 ` Hirokazu Takahashi
2002-10-14 14:12 ` Andrew Theurer
2002-10-16 3:44 ` Neil Brown
2002-10-16 4:31 ` David S. Miller
2002-10-16 15:04 ` Andrew Theurer
2002-10-17 2:03 ` [NFS] " Andrew Theurer
2002-10-17 2:31 ` Hirokazu Takahashi
2002-10-17 13:16 ` Andrew Theurer [this message]
2002-10-17 13:16 ` Andrew Theurer
2002-10-17 13:26 ` Hirokazu Takahashi
2002-10-17 13:26 ` [NFS] " Hirokazu Takahashi
2002-10-17 14:10 ` Andrew Theurer
2002-10-17 16:26 ` Hirokazu Takahashi
2002-10-17 16:26 ` [NFS] " Hirokazu Takahashi
2002-10-18 5:38 ` Trond Myklebust
2002-10-18 7:19 ` Hirokazu Takahashi
2002-10-18 15:12 ` Andrew Theurer
2002-10-18 15:12 ` [NFS] " Andrew Theurer
2002-10-19 20:34 ` Hirokazu Takahashi
2002-10-19 20:34 ` [NFS] " Hirokazu Takahashi
2002-10-22 21:16 ` Andrew Theurer
2002-10-22 21:16 ` [NFS] " Andrew Theurer
2002-10-23 9:29 ` Hirokazu Takahashi
2002-10-24 15:32 ` Andrew Theurer
2002-10-27 11:10 ` Hirokazu Takahashi
2002-10-16 11:09 ` Hirokazu Takahashi
2002-10-16 17:02 ` kaza
2002-10-17 4:36 ` rddunlap
2002-10-18 13:11 ` [PATCH] zerocopy NFS for 2.5.43 Hirokazu Takahashi
2002-10-23 1:18 ` Neil Brown
2002-10-23 3:53 ` Hirokazu Takahashi
2002-10-23 5:40 ` Hirokazu Takahashi
2002-10-23 6:03 ` Neil Brown
2002-10-23 22:35 ` Hirokazu Takahashi
2002-10-23 6:10 ` Neil Brown
2002-10-23 7:08 ` Hirokazu Takahashi
2002-10-23 15:23 ` Trond Myklebust
2002-10-23 21:50 ` Hirokazu Takahashi
2002-10-23 23:55 ` Trond Myklebust
2002-10-24 1:33 ` Hirokazu Takahashi
2002-10-27 10:39 ` Hirokazu Takahashi
2002-10-28 16:31 ` Trond Myklebust
2002-10-28 23:39 ` Hirokazu Takahashi
2002-10-29 6:36 ` Hirokazu Takahashi
2002-10-29 15:09 ` Trond Myklebust
2002-10-29 16:27 ` Hirokazu Takahashi
2002-10-29 16:49 ` Trond Myklebust
2002-10-30 3:18 ` Hirokazu Takahashi
2002-10-25 9:52 ` Hirokazu Takahashi
2002-10-25 12:41 ` Neil Brown
2002-10-26 3:11 ` Hirokazu Takahashi
2002-10-26 3:46 ` Benjamin LaHaise
2002-10-27 22:46 ` Neil Brown
2002-10-30 23:29 ` Hirokazu Takahashi
2002-10-30 23:53 ` Neil Brown
2002-10-31 2:06 ` Hirokazu Takahashi
2002-10-31 15:40 ` Hirokazu Takahashi
2002-10-31 16:56 ` Hirokazu Takahashi
2002-11-01 1:10 ` Neil Brown
2002-11-04 21:13 ` Andrew Theurer
2002-11-01 0:54 ` Neil Brown
2002-11-01 1:39 ` Hirokazu Takahashi
2002-11-01 3:41 ` Hirokazu Takahashi
2002-11-01 4:20 ` Neil Brown
2002-11-01 5:07 ` Hirokazu Takahashi
2002-10-25 17:23 ` Trond Myklebust
2002-10-26 3:26 ` Hirokazu Takahashi
[not found] <3D89176B.40FFD09B@digeo.com.suse.lists.linux.kernel>
[not found] ` <20020919.221513.28808421.taka@valinux.co.jp.suse.lists.linux.kernel>
[not found] ` <3D8A36A5.846D806@digeo.com.suse.lists.linux.kernel>
2002-09-20 1:00 ` Re: [PATCH] zerocopy NFS for 2.5.36 Andi Kleen
2002-09-20 1:09 ` Andrew Morton
2002-09-20 1:23 ` Andi Kleen
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to='003301c275e1$0bf76810$2a060e09@beavis' \
--to=habanero@us.ibm.com \
--cc=davem@redhat.com \
--cc=linux-kernel@vger.kernel.org \
--cc=neilb@cse.unsw.edu.au \
--cc=nfs@lists.sourceforge.net \
--cc=taka@valinux.co.jp \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.