[rhn-users] kswapd problems

[Doctor Khumalo]

I have two servers running RH Enterprise Linux (release) 3 with the latest 
up2date patches running on kernel 2.4.21-20.ELsmp.  These are Dell PowerEdge 
1750 server with external Powervault 220s RAID. Both servers are 
experiencing latency when running basic I/O commands such as scp or rsync. 
These appear to cause kswapd to rack up an enormous amount of processor 
resources. For example, yesterday morning, I began an scp session on each 
server, transferring a directory with a size of 300 GB. This morning I came 
in and found that the load average was over 500 on each server and that 
kswapd had a CPU percentage of 198.

This has happened repeatedly over the past month.  To replicate this, I log 
on to the first server and scp the files and directories from a file system 
on the second server.  This is reciprocated by running an scp process for 
files and directories from a file system the first server.  Both file 
systems are on /dev/sdb, an external RAID box, and are over 300 GB. Both 
servers are on the same Gigabit network.

The files transfer at approximately 15 GB an hour until about eight hours 
into the job when it begins to slow down to about 4-5 GB an hour and the 
load average increases over 200. It is at this point that kswapd begins to 
go through the roof while ssh and scp use relatively little resources 
(according to top and ps wauxg).

After I kill the scp job, about 20 hours into it, kswapd continues to use 
the same amount of resources.

I have done this CPU hyper threading enabled and disabled with no 
improvement.

Anyone ever experienced this before?

_________________________________________________________________
Is your PC infected? Get a FREE online computer virus scan from McAfee® 
Security. http://clinic.mcafee.com/clinic/ibuy/campaign.asp?cid=3963