<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=us-ascii">
<TITLE>RE: [Linux-cluster] GFS performance</TITLE>
<META content="MSHTML 6.00.2900.2995" name=GENERATOR></HEAD>
<BODY>
<DIV dir=ltr align=left><SPAN class=823003817-18122006><FONT face=Arial
color=#0000ff size=2>Well it appears you nailed it Bob. Swapping out the node1
inteconnect/heartbeat cable seemed to do the trick. I didnt look at that before
since I made/tested cables and had no issues with the OS. Thanx
much!!!</FONT></SPAN></DIV><BR>
<DIV class=OutlookMessageHeader lang=en-us dir=ltr align=left>
<HR tabIndex=-1>
<FONT face=Tahoma size=2><B>From:</B> linux-cluster-bounces@redhat.com
[mailto:linux-cluster-bounces@redhat.com] <B>On Behalf Of </B>Frazier, Darrell
USA CRC (Contractor)<BR><B>Sent:</B> Monday, December 18, 2006 8:37
AM<BR><B>To:</B> 'linux clustering'<BR><B>Subject:</B> RE: [Linux-cluster] GFS
performance<BR></FONT><BR></DIV>
<DIV></DIV>
<P><FONT size=2>Thanx a lot Bob for the info. I am in the process of swapping
things out to see if that changes things. I am wondering however why performance
is even on the OS side (dd, cp, etc, to the GFS filesystem), but not with
Clusterware. Interesting. </FONT></P>
<P><FONT size=2>-----Original Message-----</FONT> <BR><FONT size=2>From:
linux-cluster-bounces@redhat.com [<A
href="mailto:linux-cluster-bounces@redhat.com">mailto:linux-cluster-bounces@redhat.com</A>]
On Behalf Of Robert Peterson</FONT> <BR><FONT size=2>Sent: Friday, December 15,
2006 3:33 PM</FONT> <BR><FONT size=2>To: linux clustering</FONT> <BR><FONT
size=2>Subject: Re: [Linux-cluster] GFS performance</FONT> </P>
<P><FONT size=2>Frazier, Darrell USA CRC (Contractor) wrote:</FONT> <BR><FONT
size=2>></FONT> <BR><FONT size=2>> Hello guru's,</FONT> <BR><FONT
size=2>></FONT> <BR><FONT size=2>> Interesting problem I am hoping someone
on this forum has seen before, </FONT><BR><FONT size=2>> and can give me
pointers to what may be wrong. Here is the setup:</FONT> <BR><FONT
size=2>></FONT> <BR><FONT size=2>> 2 HP DL380 G4 systems with
RHEL4U4</FONT> <BR><FONT size=2>> 1 unmanaged switch (for Cluster
interconnect and Oracle RAC</FONT> <BR><FONT size=2>> interconnect.)</FONT>
<BR><FONT size=2>> Public network for Oracle NET and normal network
traffic</FONT> <BR><FONT size=2>> RHCS4 using DLM locking protocol (three
servers to keep GULM lock info </FONT><BR><FONT size=2>> for a two node RAC
seemed pretty wasteful)</FONT> <BR><FONT size=2>> GFS6.1</FONT> <BR><FONT
size=2>> Oracle Clusterware (latest)</FONT> <BR><FONT size=2>> Oracle
Database 10gR2</FONT> <BR><FONT size=2>> Fiber-Channel shared storage</FONT>
<BR><FONT size=2>></FONT> <BR><FONT size=2>> Here is my issue:</FONT>
<BR><FONT size=2>></FONT> <BR><FONT size=2>> I have set up a two-node
Oracle 10gR2 RAC system with RHCS/GFS (OCFS2 </FONT><BR><FONT size=2>> was
such a disappointment), and Oracle Clusterware. Everything on the
</FONT><BR><FONT size=2>> OS cluster level is good and fine.</FONT> <BR><FONT
size=2>></FONT> <BR><FONT size=2>> Shortly after setting up the cluster
and handing it over to the DBAs </FONT><BR><FONT size=2>> to install
Clusterware and Oracle RAC. They come to me saying that </FONT><BR><FONT
size=2>> they have uneven performance between the nodes. (Node one does
</FONT><BR><FONT size=2>> adatabase import in an hour whereas the same import
on Node2 takes 10</FONT> <BR><FONT size=2>> hours)</FONT> <BR><FONT
size=2>></FONT> <BR><FONT size=2>> I have been doing everything I can do
on my side using various tools </FONT><BR><FONT size=2>> to try to isolate
the issue (protocol analysis, iostat, strace, dd to </FONT><BR><FONT size=2>>
do writes from both nodes to GFS, etc.) and I have been unable to
</FONT><BR><FONT size=2>> isolate the issue as performance is quite even on
an OS level.</FONT> <BR><FONT size=2>></FONT> <BR><FONT size=2>> The
cluster configuration has been ruled out by Redhat support as an
</FONT><BR><FONT size=2>> issue. (Yay for me and Redhat!)</FONT> <BR><FONT
size=2>></FONT> <BR><FONT size=2>> I am hoping someone here has run into
issues using Oracle Clusterware </FONT><BR><FONT size=2>> on RHCS/GFS.</FONT>
<BR><FONT size=2>></FONT> <BR><FONT size=2>> Thanx in advance</FONT>
<BR><FONT size=2>></FONT> <BR><FONT size=2>> *Darrell J. Frazier*</FONT>
<BR><FONT size=2>></FONT> <BR><FONT size=2>Hi Darrell,</FONT> </P>
<P><FONT size=2>Well, it's possible you're running into this:</FONT> </P>
<P><FONT size=2><A href="http://sources.redhat.com/cluster/faq.html#gfs_speed1"
target=_blank>http://sources.redhat.com/cluster/faq.html#gfs_speed1</A></FONT>
</P>
<P><FONT size=2>To rule that out, perhaps you could do the same sequence of
events after a complete cluster reboot, for each node. It would be
interesting to know if the speed factor changes.</FONT></P>
<P><FONT size=2>It could be a number of other things, too, including
hardware. You could try swapping cables and ports on the Ethernet and also
the shared storage. By the way, I'm in the process of adding a new "GFS
performance tuning" question to the cluster FAQ that may or may not help.
It should appear in the FAQ soon, pending review by some of the developers here
(I don't want to post any misinformation, so I'm having them review
it).</FONT></P>
<P><FONT size=2>Regards,</FONT> </P>
<P><FONT size=2>Bob Peterson</FONT> <BR><FONT size=2>Red Hat Cluster
Suite</FONT> </P>
<P><FONT size=2>--</FONT> <BR><FONT size=2>Linux-cluster mailing list</FONT>
<BR><FONT size=2>Linux-cluster@redhat.com</FONT> <BR><FONT size=2><A
href="https://www.redhat.com/mailman/listinfo/linux-cluster"
target=_blank>https://www.redhat.com/mailman/listinfo/linux-cluster</A></FONT>
</P></BODY></HTML>