HPE 9000 and HPE e3000 Servers
cancel
Showing results for 
Search instead for 
Did you mean: 

Memory/CPU Extended Excerise

 
SOLVED
Go to solution
TYP3R
Frequent Advisor

Memory/CPU Extended Excerise

Hi

Is there a way/script that can excerise memory/CPU continuously ? I know i can excerise using stm but that only gives a good 10min excerise.. I need excerise these memory and CPU continuosly for over night incase it fails over, Then we can install it on a production server.

Any kind of help would be grateful

Will
4 REPLIES 4
mits
Respected Contributor
Solution

Re: Memory/CPU Extended Excerise

You can change it with the excersciser option. Use eop or exeroptions.

cstm>eop
-- Exercise Options --
Refer to Tool specific help for the effect of these option settings.

Execution Control:
* 1 Execution Time Limit, in Minutes (time)
2 Loop Continuously (continuously)
Enter selection: [time] 2
Bill Hassell
Honored Contributor

Re: Memory/CPU Extended Excerise

Here is a script that you can run and specify the number of minutes on the command line (1 min is the default). Start it with the number 1440 for a 24 hour run. It is particularly useful because it will use all memory including virtual memory and all CPUs:

#!/usr/bin/sh

##################
# #
# cpumemBusy #
# #
##################

# Usage: cpumemBusy [ mins ]
# default mins=1

[ $# -gt 0 ] && MINS=$1 || MINS=1
echo "\nExercise CPU and memory for $MINS minute$S\n"
echo "scl qualifier cpu;\
scl qualifier memory;\
eop time $MINS maxcoverage;\
exc;\
wait;\
eal;" | /usr/sbin/cstm


Bill Hassell, sysadmin
Highlighted
TYP3R
Frequent Advisor

Re: Memory/CPU Extended Excerise

Thank Guys for that...

Bill does it give any type of output indicating that the cpu or memory have failed ?

Will
Bill Hassell
Honored Contributor

Re: Memory/CPU Extended Excerise

Yes, cstm exerciser will give you a lot of details. Just run this a couple of minutes and you'll see lots of details. Now if a CPU or memory fails, cstm is just an application so you will get all the nasty kernel messages (including a crash) when a component fails. It's not like an offline diagnostic which can pinpoint a bad CPU or bad memory. These are logged just as they would be in a running system. Always look at syslog after a long run.


Bill Hassell, sysadmin