HMC and replaced RAM DIMM. Still have Deconfigured RAM 1

[jpor2003]

Hi Gurus,

I have been currently tasked looking into an issue with a Rack of LPARs showing deconfigured RAM. Although the IBM engineer has replaced the RAM DIMM. When I look on the HMC against the rack, I still have 10GB RAM showing as de-configured. Looking into this I have found an option under 'Service Applications' - 'Service Focal Point' - 'Repair Servicable Event'. But when I come to run through the repair procedure, I am advised I need to power of all the partitions. Before I go ahead and get agreement to do this. Is this the right option to sort out this missing RAM problem? The LPARs are running on an P5 570+ with AIX 5.3 TL04 CSP.

Thnaks.

 

[madmurdock]

I would suggest you follow the advice to power off all lpars and the managed system. It would no surprise to me, if its the only way to detect new or changed RAM DIMMs.

As this is absolutey OS-independently, the TL of your LPARs is not important.

Hth

mad

Advanced Interactive eXecutable

 

[khalidaaa]

Did the IBM engineer replaced the RAM while the LPARs running using that RAM?

I think this shouldn't be the case!

I remember when i asked my local IBM engineer he said that the system should go off for this action to be done! adding more RAM i mean!

Regards,
Khalid

 

[madmurdock]

If its so, like Khalid says, then the lpars would not even run but crashed, or not?

I think jpor2003 did right to running "Repair Servicable Event" and the instructions are mostly right..

mad

Advanced Interactive eXecutable

 

[jpor2003]

Thanks for the advice guys. It was one of my colleagues who performed the neccessaries, and had an IBM engineer in to replace the RAM. So I am hoping that the whole thing was powered off before hand. What I suspect though is that this deconfigured RAM is a left over from the original error, and thus I am looking into how this can be cleared down, and allow for this missing RAM to be used.
So I take it 'Reapir Servicable Event' is the right way to go then?
Just don't relish the idea of getting Live machines powered off for nothing.

 

[madmurdock]

Are there any alternatives ? Honestly, if the RAM is not detected after the change, I would follow the instructions. I cant see another way.

Hth

mad

Advanced Interactive eXecutable

 

[rzs0502]

I seem to re-call doing this (for a processor) using the ASM Web Interface.

Logon to the ASM using the 'admin' username
Go to System Configuration
Go to Hardware Deconfiguration

Select the unit and continue. It should show you the status of all the memory where you can select 'Configure' to enable it.

The IBM engineer should have done this.



"If you always do what you've always done, you will always be where you've always been."

 

[p5wizard]

If you/IBM added RAM with the system running, you probably had CuOD memory in your server which was activated - this can be done on the fly.

Physically adding memory to a p5-570 can only be done with the system powered off, because the CPU boxes need to come out: CPU-flexcable and SP-flexcable need to come off, slide out CPU-box, open it up, take out DIMM dummies, insert real DIMMs, close CPU-box, reinsert CPU-box, repeat for other CPU boxes - I don't see that happening while anything is running on the server...

For deconfigured memory to get configured again, you need a full server downtime, not just shutdown the LPARs, but also a poweroff of the server. And your IBM CE needs a special password to log in to ASMI - with the admin password you have, you can see deconfig'd memory but you can't un-deconfig it.


HTH,

p5wizard

 

[jpor2003]

rzs0502. What's the ASM? I currently use a product called websm to access the HMC. But I cannot see an option in relation to 'System Configuration'.
Please advise.

 

[hfaix]

Assuming no CuOD, I recommend you follow the steps rzs0502 mentions. I've done this processor successfully.

With that said, you're safest best is to re-open the original pmr and check it out. The Engineer (CE) should have done this.

 

[drencep]

Hi,

I hope it's okay to post a similar issue on this thread. I need to upgrade a p570 FW to 240_284 and going thru the procedure, I found out that a 2GB memory is deconfigured and it needs to be resolved before proceeding with the upgrade. Now, as p5wizard said, as user admin you can see the deconfigured memory but won't be able to configure it back. And as I understand, you need to login with a special IBM account. Is there a workaround or other ways to do this?

TIA!

 

[p5wizard]

To un-deconfig hardware, you need to call IBM and either let them calculate a celogin password for you (valid on a specific date), or get a CE (customer engineer) on site to un-deconfig it for you.

As stated above:
The userid admin in ASMI can show you deconfig'd hardware, but it won't let you alter it.

The userid celogin in ASMI CAN un-deconfig hardware.


HTH,

p5wizard

 

[p5wizard]

Of course, depending on the cause of the deconfiguration, It may well be that the DIMM needs replacing. Either way - call IBM...


HTH,

p5wizard

 

[drencep]

Thanks for your inputs. If I replace the system deconfigured DIMMS, will they come back as configured or will they stay as deconfigured.

 

[albalosr]

i believe there is an issue with the p570 for having deconfigured RAM. this has been an outstanding issue so i suggest you move to the latest server firmware if IBM has already replaced the RAM. the celogin is needed to activate the deconfigured ones. the IBM SSR should do this for you.