From: "Lever, Charles" Subject: RE: quick connection then wait but for what? Date: Mon, 29 Mar 2004 06:09:07 -0800 Sender: nfs-admin@lists.sourceforge.net Message-ID: <482A3FA0050D21419C269D13989C61130435DDFA@lavender-fe.eng.netapp.com> Mime-Version: 1.0 Content-Type: multipart/alternative; boundary="----_=_NextPart_001_01C41597.663B779F" Cc: Return-path: Received: from sc8-sf-mx1-b.sourceforge.net ([10.3.1.11] helo=sc8-sf-mx1.sourceforge.net) by sc8-sf-list2.sourceforge.net with esmtp (Exim 4.30) id 1B7xRz-00079h-5G for nfs@lists.sourceforge.net; Mon, 29 Mar 2004 06:09:15 -0800 Received: from mx01.netapp.com ([198.95.226.53]) by sc8-sf-mx1.sourceforge.net with esmtp (Exim 4.30) id 1B7xRy-0005oG-Rn for nfs@lists.sourceforge.net; Mon, 29 Mar 2004 06:09:14 -0800 To: "beooi alex" Errors-To: nfs-admin@lists.sourceforge.net List-Unsubscribe: , List-Id: Discussion of NFS under Linux development, interoperability, and testing. List-Post: List-Help: List-Subscribe: , List-Archive: This is a multi-part message in MIME format. ------_=_NextPart_001_01C41597.663B779F Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable hi alex- =20 can you share with us a network trace captured during a data transfer that shows this performance anomaly? =20 -----Original Message----- From: beooi alex [mailto:beooia@yahoo.fr]=20 Sent: Sunday, March 28, 2004 5:00 AM To: nfs@lists.sourceforge.net Subject: [NFS] quick connection then wait but for what? hello, first of all excuse me for my poor english :) this is our lvs cluster: director | switch | ---------------------------------------------- | | web/mailserver webmailserver | | ---------------------------------------------- | switch | ---------------------------------------------- | | fileserver databaseserver we have a strange problem: we don't use nor identd nor hostnamelookup but we have those symptoms... when you request a page, the result is the same with a 1ko page or a 10 = mo one, you see connection, opening, then you wait between 1 and 3 seconds, then everything appear very fast we didn't have this problem before we "clustered" our installation and dedicated a server for the files. we add a smart array 641 controller with 4 discs (raid10) for the = fileserver. we tried quite all combinaisons of rsize, wsize, tcp, udp, version 2; = version 3. we don't understand how this cluster can respond slower than before. (before we just used one server) can it be the default stripe size of the file server (128 ko for = raid10)? ...? thanks a lot in advance for your responses/help. :) alex. _____ =20 Yahoo! Mail : votre e-mail personnel et gratuit qui vous suit partout !=20 Cr=E9ez = votre Yahoo! Mail=20 Dialoguez en direct avec vos amis gr=E2ce =E0 Yahoo! = Messenger ! ------_=_NextPart_001_01C41597.663B779F Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Message
hi=20 alex-
 
can=20 you share with us a network trace captured during a data = transfer
that=20 shows this performance anomaly?
 
-----Original Message-----
From: = beooi alex=20 [mailto:beooia@yahoo.fr]
Sent: Sunday, March 28, 2004 5:00=20 AM
To: nfs@lists.sourceforge.net
Subject: [NFS] = quick=20 connection then wait but for what?

hello,

first of all excuse me for my poor english = :)

this=20 is our lvs=20 = cluster:

         &nb= sp;           &nbs= p;      =20 = director
          &= nbsp;           &n= bsp;          =20 = |
           &n= bsp;           &nb= sp;     =20 = switch
          &nb= sp;           &nbs= p;          =20 = |
           =20 = ----------------------------------------------
    = ;      =20 = |            =             &= nbsp;           &n= bsp;        =20 = |
web/mailserver         =             &= nbsp;   =20 = webmailserver
         &n= bsp;=20 = |            =             &= nbsp;           &n= bsp;        =20 = |
           =20 = ----------------------------------------------
    = ;            =             &= nbsp;    =20 = |
           &n= bsp;           &nb= sp;     =20 = switch
          &nb= sp;           &nbs= p;          =20 = |
           =20 = ----------------------------------------------
    = ;      =20 = |            =             &= nbsp;           &n= bsp;        =20 |
    =20 = fileserver          &nb= sp;           &nbs= p;      =20 databaseserver

we have a strange problem:
we don't use nor = identd=20 nor hostnamelookup but we have those symptoms...

when you = request a=20 page, the result is the same with a 1ko page or a 10 mo one,
you = see=20 connection, opening, then you wait between 1 and 3 seconds,
then = everything=20 appear very fast

we didn't have this problem before we = "clustered" our=20 installation
and dedicated a server for the files.
we add a smart array 641 = controller with 4 discs (raid10) for the fileserver.
we tried quite all combinaisons of rsize, wsize, tcp, udp, = version 2;=20 version 3.

we don't understand how this cluster can respond slower than=20 before.
(before we just used one server)

can it be the default = stripe size=20 of the file server (128 ko for raid10)?
...?

thanks a lot in = advance=20 for your responses/help. :)

alex.

Yahoo! Mail : votre e-mail personnel et gratuit qui vous suit partout = !
Cr=E9ez=20 votre Yahoo! Mail

Dialoguez en direct avec vos amis = gr=E2ce =E0 Yahoo!=20 Messenger ! ------_=_NextPart_001_01C41597.663B779F-- ------------------------------------------------------- This SF.Net email is sponsored by: IBM Linux Tutorials Free Linux tutorial presented by Daniel Robbins, President and CEO of GenToo technologies. Learn everything from fundamentals to system administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click _______________________________________________ NFS maillist - NFS@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/nfs