All of lore.kernel.org
 help / color / mirror / Atom feed
From: Eric Whiting <ewhiting@amis.com>
To: "Lever, Charles" <Charles.Lever@netapp.com>
Cc: nfs@lists.sourceforge.net
Subject: Re: 2.6.0 -- nfs client runs 15x faster if client app runs as root.
Date: Wed, 07 Jan 2004 14:00:33 -0700	[thread overview]
Message-ID: <3FFC7371.6279A9F6@amis.com> (raw)
In-Reply-To: 482A3FA0050D21419C269D13989C6113020AC9B6@lavender-fe.eng.netapp.com

"Lever, Charles" wrote:
> 
> are you writing to a set-uid file?  in the normal user case,
> the SETATTR is required by Posix to unset the set-* bits on
> the file.

It is a 'bonnie' run.. This test simply creates a new file and writes/reads the
file.

I don't see any suid bits on these files.

I tried the same tests on a Solaris 9 nfs server. Same results.


eric


> > -----Original Message-----
> > From: Eric Whiting [mailto:ewhiting@amis.com]
> > Sent: Wednesday, January 07, 2004 3:14 PM
> > To: nfs@lists.sourceforge.net
> > Subject: [NFS] 2.6.0 -- nfs client runs 15x faster if client
> > app runs as
> > root.
> >
> >
> > My 2.6.0 nfs clients to a netapps 940 NFS server run very
> > differently depending
> > on who the user is. Both tcp and udp mounts seem to show the
> > same behavior. This
> > is very apparent when doing the bonnie putc() tests.
> >
> > nfsstat and tcpdump show a lot of extra setattr calls when
> > running as a normal
> > user. I assume this is the reason for the slowdown.
> >
> > Is there a fix? Or is this the way it is supposed to be?
> >
> > This is not how this same hardware acts in 2.4.23.
> >
> > Thanks,
> > eric
> >
> >
> >
> > Running as a 'user' doing bonnie putc()
> > Writing with putc()...  done:    802 kB/s  10.6 %CPU
> >
> > 13:06:15.841823 172.16.17.176.3073181638 > 172.16.32.90.2049:
> > 144 setattr [|nfs]
> > (DF)
> > 13:06:15.842024 172.16.32.90.2049 > 172.16.17.176.3073181638:
> > reply ok 144
> > setattr [|nfs]
> > 13:06:15.842361 172.16.17.176.3089958854 > 172.16.32.90.2049:
> > 648 write [|nfs]
> > (DF)
> > 13:06:15.842611 172.16.32.90.2049 > 172.16.17.176.3089958854:
> > reply ok 160 write
> > [|nfs]
> > 13:06:15.842880 172.16.17.176.3106736070 > 172.16.32.90.2049:
> > 144 setattr [|nfs]
> > (DF)
> > 13:06:15.843079 172.16.32.90.2049 > 172.16.17.176.3106736070:
> > reply ok 144
> > setattr [|nfs]
> > 13:06:15.843506 172.16.17.176.3123513286 > 172.16.32.90.2049:
> > 648 write [|nfs]
> > (DF)
> > 13:06:15.843783 172.16.32.90.2049 > 172.16.17.176.3123513286:
> > reply ok 160 write
> > [|nfs]
> > 13:06:15.844069 172.16.17.176.3140290502 > 172.16.32.90.2049:
> > 144 setattr [|nfs]
> > (DF)
> > 13:06:15.844253 172.16.32.90.2049 > 172.16.17.176.3140290502:
> > reply ok 144
> > setattr [|nfs]
> > 13:06:15.844604 172.16.17.176.3157067718 > 172.16.32.90.2049:
> > 648 write [|nfs]
> > (DF)
> >
> > Running as root on the client box doing a bonnie putc()
> > Writing with putc()...         done:  11625 kB/s  70.5 %CPU
> >
> > 13:04:34.530773 172.16.32.90.2049 > 172.16.17.176.2934638278:
> > reply ok 160 write
> > [|nfs]
> > 13:04:34.533983 172.16.17.176.2951415494 > 172.16.32.90.2049:
> > 1472 write [|nfs]
> > (frag 58503:1480@0+)
> > 13:04:34.533995 172.16.17.176 > 172.16.32.90: udp (frag
> > 58503:1480@1480+)
> > 13:04:34.533999 172.16.17.176 > 172.16.32.90: udp (frag
> > 58503:1480@2960+)
> > 13:04:34.534003 172.16.17.176 > 172.16.32.90: udp (frag
> > 58503:1480@4440+)
> > 13:04:34.534007 172.16.17.176 > 172.16.32.90: udp (frag
> > 58503:1480@5920+)
> > 13:04:34.534010 172.16.17.176 > 172.16.32.90: udp (frag
> > 58503:1480@7400+)
> > 13:04:34.534014 172.16.17.176 > 172.16.32.90: udp (frag
> > 58503:1480@8880+)
> > 13:04:34.534018 172.16.17.176 > 172.16.32.90: udp (frag
> > 58503:1480@10360+)
> > 13:04:34.534021 172.16.17.176 > 172.16.32.90: udp (frag
> > 58503:1480@11840+)
> > 13:04:34.534024 172.16.17.176 > 172.16.32.90: udp (frag
> > 58503:1480@13320+)
> > 13:04:34.534028 172.16.17.176 > 172.16.32.90: udp (frag
> > 58503:1480@14800+)
> > 13:04:34.534032 172.16.17.176 > 172.16.32.90: udp (frag
> > 58503:1480@16280+)
> > 13:04:34.534035 172.16.17.176 > 172.16.32.90: udp (frag
> > 58503:1480@17760+)
> > 13:04:34.534039 172.16.17.176 > 172.16.32.90: udp (frag
> > 58503:1480@19240+)
> > 13:04:34.534042 172.16.17.176 > 172.16.32.90: udp (frag
> > 58503:1480@20720+)
> > 13:04:34.534045 172.16.17.176 > 172.16.32.90: udp (frag
> > 58503:1480@22200+)
> > 13:04:34.534049 172.16.17.176 > 172.16.32.90: udp (frag
> > 58503:1480@23680+)
> > 13:04:34.534053 172.16.17.176 > 172.16.32.90: udp (frag
> > 58503:1480@25160+)
> > 13:04:34.534056 172.16.17.176 > 172.16.32.90: udp (frag
> > 58503:1480@26640+)
> > 13:04:34.534060 172.16.17.176 > 172.16.32.90: udp (frag
> > 58503:1480@28120+)
> > 13:04:34.534063 172.16.17.176 > 172.16.32.90: udp (frag
> > 58503:1480@29600+)
> > 13:04:34.534067 172.16.17.176 > 172.16.32.90: udp (frag
> > 58503:1480@31080+)
> > 13:04:34.534070 172.16.17.176 > 172.16.32.90: udp (frag
> > 58503:368@32560)
> > 13:04:34.537357 172.16.32.90.2049 > 172.16.17.176.2951415494:
> > reply ok 160 write
> > [|nfs]
> > 13:04:34.540586 172.16.17.176.2968192710 > 172.16.32.90.2049:
> > 1472 write [|nfs]
> > (frag 58504:1480@0+)
> > 13:04:34.540598 172.16.17.176 > 172.16.32.90: udp (frag
> > 58504:1480@1480+)
> > 13:04:34.540603 172.16.17.176 > 172.16.32.90: udp (frag
> > 58504:1480@2960+)
> >
> >
> > Client nfs v3:
> > null       getattr    setattr    lookup     access     readlink
> > 0       0% 8724    0% 1113998 47% 5225    0% 3084    0% 98      0%
> > read       write      create     mkdir      symlink    mknod
> > 20776   0% 1213403 51% 237     0% 1       0% 3       0% 0       0%
> > remove     rmdir      rename     link       readdir    readdirplus
> > 206     0% 0       0% 95      0% 24      0% 48      0% 416     0%
> > fsstat     fsinfo     pathconf   commit
> > 118     0% 10      0% 0       0% 0       0%
> >
> >
> > -------------------------------------------------------
> > This SF.net email is sponsored by: Perforce Software.
> > Perforce is the Fast Software Configuration Management System offering
> > advanced branching capabilities and atomic changes on 50+ platforms.
> > Free Eval! http://www.perforce.com/perforce/loadprog.html
> > _______________________________________________
> > NFS maillist  -  NFS@lists.sourceforge.net
> > https://lists.sourceforge.net/lists/listinfo/nfs
> >


-------------------------------------------------------
This SF.net email is sponsored by: Perforce Software.
Perforce is the Fast Software Configuration Management System offering
advanced branching capabilities and atomic changes on 50+ platforms.
Free Eval! http://www.perforce.com/perforce/loadprog.html
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

  reply	other threads:[~2004-01-07 21:00 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2004-01-07 20:46 2.6.0 -- nfs client runs 15x faster if client app runs as root Lever, Charles
2004-01-07 21:00 ` Eric Whiting [this message]
  -- strict thread matches above, loose matches on Subject: below --
2004-01-08 15:12 trond.myklebust
2004-01-08 15:23 ` trond.myklebust
2004-01-07 21:54 trond.myklebust
2004-01-08 14:15 ` Gaël Roualland
2004-01-08 15:12   ` Eric Whiting
2004-01-07 20:13 Eric Whiting

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=3FFC7371.6279A9F6@amis.com \
    --to=ewhiting@amis.com \
    --cc=Charles.Lever@netapp.com \
    --cc=nfs@lists.sourceforge.net \
    /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.