The system takes over

jason andrade jason at dstc.edu.au
Wed Nov 27 17:50:00 CST 2002


On Thu, 28 Nov 2002, Norman Gaywood wrote:

> OK. I have no H/W raid controller.  I've attached the syslog startup to
> this message which I think should have all the numbers you need. Here
> is /etc/fstab:

Hi Norman,

At first guess there is definitely a problem with the system which might
not even be raid or filesystem related.  as andrew pointed out, there is
definitely an issue when you have 99% of cpu time being spent in system
as the kernel is obviously blocked (or looped) on something there.

This is an example of a relatively busy NFS server which is pushing out
100Mbit 24/7 using 64 nfsds as well as running several rsync processes
and a software raid5 array.  it is only a PE2400 w/ 2 cpus and 1G of ram.

CPU0 states:  0.0% user, 65.1% system,  0.0% nice, 34.3% idle
CPU1 states:  0.0% user, 67.2% system,  0.0% nice, 32.2% idle


What is the behaviour of the system like without running the rsync process
over a period of time ?  what does /proc/interrupts show you - are they
balanced across cpus ?

Are you running the latest errata kernel ?

> md6 : active raid5 sdg1[4] sdf1[3] sde1[2] sdd1[1] sdc1[0]
>       215045760 blocks level 5, 64k chunk, algorithm 0 [4/4] [UUUU]
>       [========>............]  resync = 44.4% (31848552/71681920)
> finish=64.3min speed=10313K/sec
> unused devices: <none>

you can speed this up by increasing the values in /proc/sys/dev/raid/

regards,

-jason





More information about the Linux-PowerEdge mailing list