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

9000 server hardware testing - guru's needed!

 
SOLVED
Go to solution
Steven Bucek
Frequent Advisor

9000 server hardware testing - guru's needed!

What is the best method of testing to ensure full functionality of all hardware on L and N class machines? or ANY 9000 gear for that matter. Are the self tests reliable? I have never seen anything say 'failed' on the self tests as they are scrolling by, what happens when there is a failure? I have seen bad processors that would cause the self tests to freeze up, and work fine with replaced with a working processor, but never seen 'failed' anywhere.

The ODE is almost useless unless you have a fist full of codewords, im sure money is involved in getting those!

I would like to have something that says 'X has been tested and PASSED' that I could print out. I need some proof.

Please share your thoughts or testing methods.

--To go forward, you must backup
4 REPLIES 4

Re: 9000 server hardware testing - guru's needed!

If you look in stm (xstm is probably the one to use if your not familiar with it) you will find that most devices have an 'exercise' tool that can be run against them to confirm they are functioning correctly. That seems like a good place to start...

HTH

Duncan

Accept or Kudo
Steven Bucek
Frequent Advisor

Re: 9000 server hardware testing - guru's needed!

Thanks for your response. I dont have access to the OS here, I just config the hardware and pass it along.

Seems like there should be a more simple way to test hardware than doing an install and using STM on everything..

--To go forward, you must backup
Andrew Rutter
Honored Contributor
Solution

Re: 9000 server hardware testing - guru's needed!

Hi STM is the most reliable method for testing the hardware and you can print the logs/results if needed.

Without the os its either ODE which isnt as good and needs a password from HP to use its full functionality.

Failing that if the system gets to pdc level without problem the system may well be okay. Any serious errors should stop the system booting further. You would get chassis codes and errors in the gsp logs for any failures for system board/mem/proc/psu/pmc but not for any IO cards that may be present in the system.
The system would not even get to pdc if there was a serious problem. The lights would flash on the l/n and then you would have to read the corresponding logs in the gsp to identify the problem.

This is only basic tests though and not with the system under any load or doing a full exercise.
If I want to fully test a system or component I always install hpux and use stm and do proper exercises on the mem/proc etc.

Andy


Steven Bucek
Frequent Advisor

Re: 9000 server hardware testing - guru's needed!

Ok seems the BEST way to test hardware is to put parts in an OS evniroment and test them with STM. Those of you under support contracts may have access to codewords needed in the Offline Diagnostic Enviroment.
GSP error logs are use if there are problems booting to PDC.

Thanks
--To go forward, you must backup