 | | | Another (Unrelated) IOWAIT Problem | Another (Unrelated) IOWAIT Problem 2006-02-22 - By John.B.Kelly
Back
Hello,
We're using a Poweredge 700 running RHEL AS3 to NFS share some data. The system
does not have a 3Ware SATA controller so I don't think bugzilla bug 121434
(https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=121434) is relevant. The
problem (which is causing severe NFS client performace degradation) is that
when I run top, iowait is up at around 90 to 95%. There is also audible hard
-disk thrashing. If I discontinue NFS sharing from the Poweredge, iowait goes
down to almost 0. When I restart NFS sharing, iowait stays low for an arbitrary
amount of time and then goes up to 90/95% again. There doesn't seem to be any
correlation that I can establish between any accessing of the NFS shared
directories by NFS clients and when the iowait parameter returns to 90/95%. Has
anybody seen this before? I
'm currently running kernel 2.4.21-37.0.1.EL but also saw this problem with
some 2.4.21-27 kernels. I searched for solutions on the web and as a result I
've been playing with some kernel parameters in /proc/sys/vm/inactive_clean
_percent and /proc/sys/vm/pagecache (currently set to 100 and "2 10 20"
respectively) but this has not caused any improvement. The next 2 things which
I am considering trying is installing autofs on the NFS clients (to possibly
reduce network traffic) or else install a 2.6 kernel on the NFS server but
before I do either of these, I would really appreciate any advice from any RHEL
AS3 users who have seen this problem or even who could recommend an appropriate
next step in my troubleshooting.
Thanks,
John.
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859 (See http://iso-8859.ora-code.com)-1">
<META NAME="Generator" CONTENT="MS Exchange Server version 6.5.7233.28">
<TITLE>Another (Unrelated) IOWAIT Problem</TITLE>
</HEAD>
<BODY>
<!-- Converted from text/rtf format -->
<P><FONT SIZE=2 FACE="Arial">Hello,</FONT>
</P>
<P><FONT SIZE=2 FACE="Arial">We're using a Poweredge 700 running RHEL AS3 to
NFS share some data. The system does not have a 3Ware SATA controller so I don
't think bugzilla bug 121434 (<A HREF="https://bugzilla.redhat.com/bugzilla/show
_bug.cgi?id=121434">https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=121434<
/A>) is relevant. The problem (which is causing severe NFS client performace
degradation) is that when I run top, iowait is up at around 90 to 95%. There is
also audible hard-disk thrashing. If I discontinue NFS sharing from the
Poweredge, iowait goes down to almost 0. When I restart NFS sharing, iowait
stays low for an arbitrary amount of time and then goes up to 90/95% again.
There doesn't seem to be any correlation that I can establish between any
accessing of the NFS shared directories by NFS clients and when the iowait
parameter returns to 90/95%. Has anybody seen this before? I</FONT></P>
<P><FONT SIZE=2 FACE="Arial">'m currently running kernel 2.4.21-37.0.1.EL but
also saw this problem with some 2.4.21-27 kernels. I searched for solutions on
the web and as a result I've been playing with some kernel parameters in /proc
/sys/vm/inactive_clean_percent and /proc/sys/vm/pagecache (currently set to 100
and "2 10 20" respectively) but this has not caused any improvement.
The next 2 things which I am considering trying is installing autofs on the NFS
clients (to possibly reduce network traffic) or else install a 2.6 kernel on
the NFS server but before I do either of these, I would really appreciate any
advice from any RHEL AS3 users who have seen this problem or even who could
recommend an appropriate next step in my troubleshooting. </FONT></P>
<P><FONT SIZE=2 FACE="Arial">Thanks,</FONT>
<BR><FONT SIZE=2 FACE="Arial">John.</FONT>
</P>
</BODY>
</HTML>
--
Taroon-list mailing list
Taroon-list@(protected)
https://www.redhat.com/mailman/listinfo/taroon-list
|
|
|