Need help with errpt generating an errpt entry 1

[zaxxon]

Hi,

I want to simulate a hardware error in errpt. I don't want to draw a disk or something to generate it. I found "errlogger" to produce an entry, but it's just an operator message with different class and such. What I need is a hardware error, PERM or TEMP doesn't matter.

Thanks for any ideas!

laters
zaxxon

 

[khalidaaa]

errlogger is the way to go! but if you want to simulate a real error then try for example removing a fiber channel from an active san! or an ethernet cable!

Regards,
Khalid

 

[zaxxon]

If errlogger is the way to go, I am too dumb to see, but there is no way to specify a hardware error. The man page has no command switches but only parameters for the message text. As I said, I don't want to draw any cable or disk etc.

laters
zaxxon

 

[p5wizard]

Do diag test on operator panel, give wrong answer to the tests - that will generate a HW error about operator panel.

Or pull/reseat a network cable.


HTH,

p5wizard

 

[zaxxon]

Ok, you got me. I will go and pull a network cable. Good way to test Etherchannel Backup Interface anyway. Thanks all.

laters
zaxxon

 

[DukeSSD]

p5wizard's diag test is the way to go if you can't be bothered to get out of your seat or if the box is miles away.

That is how IBM suggest you test error reporting and service agent's call home function because it generates a Perm hardware error without any disruptiion to the system.

I hope you etherchannel test is a success, or you don't try it on a live system during production hours.

 

[zaxxon]

I used diag before on different devices, but can't remember that I've ever got any questions but if the resource if free or not. I checked diag's man page and tried some of diagnostic routines but got no questions.

What options should I select to get to the point where the questions are asked?
Or do you mean some diag via hmc?

laters
zaxxon

 

[p5wizard]

On a non-HMC system:

diag
-> Advanced Diagnostics Routines
-> System Verification
-> choose "oppanel" (press enter to put a + sign in front)
-> F7
-> Enter to start test
-> answer NO to first question on "All zeroes"
That *SHOULD* enter a PERM H error in errlog...

If the system is managed by an HMC you can't run tests on device "oppanel", however instead it asks you if you want to create a test Serviceable Event so you can verify the "CallHome" functionality of your HMC.

Is that enough?



HTH,

p5wizard

 

[zaxxon]

The system is HMC managed and it asked me for the event like you said. I said "YES" and it displayed my the test message, which I accepted with "YES" to be sent too.
I checked the HMC service functions and listed all service messages from the appropriate hardware/system but they were all much older; so nothing got through from the test event I've created.
I guess I'll draw that cable.

laters
zaxxon

 

[zaxxon]

A star for explaing it to me

laters
zaxxon

 

[p5wizard]

If that event (pull network cable) doesn't get through either to ServiceFocalPoint, you have reason enough to call IBM by phone and let IBM support figure out what is wrong...

A hint though: are your server's LIC level and the HMC installation level current enough and compatible with one another? See this site:

http://www14.software.ibm.com/webapp/set2/sas/f/power5cm/supportedcodep5.html

On a side note, on the HMC there's also a possibility to generate a test Servicable Event, so you can at least make sure the HMC "CallHome" setup is good. Also you may want to enable heartbeat for the CallHome as well.



HTH,

p5wizard

 

[zaxxon]

I checked that call-out function of the HMC earlier - there is nothing configured, that might work. Our systems are not IBM RS6K, they are from Bull and I guess we didn't buy that kind of support and so it is not configured.
I will see if the cable check generate an errpt entry and an event on the assigned HMC. If that works I am pleased

laters
zaxxon

 

[zaxxon]

Since I was too lazy yet, I found a way to generate Error Report entries:

/usr/sbin/rsct/bin/fclogerr

I started it without any options and got some ugly errors on my terminal but also an errpt entry. With the right options it might be the tool I was looking for. Just wanted to let you know, if you are interessted.

laters
zaxxon

 

[p5wizard]

At first glance, fclogerr man page only talks about software errors.

Here's another option for you: if the machines are dual power, AND both powers are good (check green LEDs on the supplies, check lscfg -vp, whatever). Pull one of the power cables and then reseat it.

Machine would then respond in notifying Service Focal Point on HMC. It would probably be a good thing to test that out on non-productive machine...


HTH,

p5wizard