SCIENTIFIC-LINUX-USERS Archives

June 2008

SCIENTIFIC-LINUX-USERS@LISTSERV.FNAL.GOV

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Felix Engel <[log in to unmask]>
Reply To:
Felix Engel <[log in to unmask]>
Date:
Wed, 18 Jun 2008 18:49:05 +0200
Content-Type:
multipart/signed
Parts/Attachments:
text/plain (1207 bytes) , smime.p7s (4 kB)
Hi Jan,


Am 18.06.2008 um 18:36 schrieb Jan Schulze:
>
> I did comparisons compiling gcc on the old and new systems while  
> running
> tcpdump. I found, that the new systems performed a lot more NFS  
> lookups
> (by factor ~4). Everything else seemed normal.
>
> File transfer to the NFS server is OK, no speed problems here.  
> Tcpdump,
> however shows a lot of lines like this one, when copying files:
>   18:19:56.571906 IP bsnode02.0 > morphin.nfs: 1448 null
>
> Load on the NFS server is much higher than before. I'm a little bit  
> at a
> loss - how could I debug these issues further?
This is a bug in the kernel seemingly caused by a weird combination  
of patches in the RHEL kernel. See (in order):

http://bugs.centos.org/view.php?id=2635
https://bugzilla.redhat.com/show_bug.cgi?id=431092
https://bugzilla.redhat.com/show_bug.cgi?id=321111

We have encountered the same problems and switching to kernel  
2.6.18-53.1.13.el5.bz321111 as reported in the CentOS bug tracking  
solved the problem for us.

Kind Regards,
	Felix

-- 
   Dipl.-Ing. Felix Engel        | mail: [log in to unmask]
   fon: +49 241 80 27629         | fax: +49 241 80 28306
   RWTH Aachen University, SSS 611920, Templergraben 55, D-52056 Aachen




ATOM RSS1 RSS2