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

CPU connectors/pins shredding - A5866-69001

 
SOLVED
Go to solution
nibble
Super Advisor

CPU connectors/pins shredding - A5866-69001

Hi everyone,
Previously, ive posted a thread about an N4000 (A3639B) that has 8 CPUs and keeps on rebooting. CPU part number is A5866-69001. Answers lead me to the trim card installation which i did (thanks to all who replied). the machine stopped rebooting for 30 days, but just recently it started again and even worse, giving only 5 mins interval. what i did was temporarily remove 2 cpus, especially the ones starting the hpmc panic and it was now up and under observation. I closely check the removed CPUs and to my disappointment, some of the rubber like pins from the chip were shredded, thus they might not be properly connected while installed and might have caused the recent panics. Im planning to order a new set of cpus, but i need to have a good justification for it.
my questions are:
1. assuming that the cpu's were installed with those shredded pins, why we're they running for a period of time?
2. does it mean that they can ran even with some of the pins not in contact?
3. are those really pins or am i just mistaken em for pins?

Ive attached a photo of it.
Thanks.
7 REPLIES 7
Steven E. Protter
Exalted Contributor

Re: CPU connectors/pins shredding - A5866-69001

Shalom,

You are going to get a HPMC every time you pull a CPU.

I would assume there is a hardware problem and you want to get a CPU with good pins, even if that costs some money.

SEP
Steven E Protter
Owner of ISN Corporation
http://isnamerica.com
http://hpuxconsulting.com
Sponsor: http://hpux.ws
Twitter: http://twitter.com/hpuxlinux
Founder http://newdatacloud.com
Matti_Kurkela
Honored Contributor
Solution

Re: CPU connectors/pins shredding - A5866-69001

The brown plastic part with the rubber-like pins is just a holder: the actual CPU is inside it, held against the bottom of the heat sink. The holder can be detached from the CPU heat sink. I think it is keyed so that it will install only one way.

When the CPU is in use, the rubber "pins" are subjected to compression and lots of heat, so they will dry and harden, and may easily be damaged when an used CPU is removed.

A HP engineer once said that the holder part should be inspected carefully whenever a CPU is re-installed: it's essentially "disposable with a possibility to re-use if it's in good condition".

If necessary, the holder part can be replaced: HP engineers may have those parts in their toolkit for L and N class servers' CPU installation jobs.

You should also inspect the CPU socket: if there are any crumbs from the broken rubber "pins", you need to get them cleaned away. They are electrically conductive and may cause faults if left between the CPU and the socket. Don't touch the gold contacts with your fingers: the natural oil/grease of human skin may be enough to cause problems.

Damaged or missing rubber "pins" are a definite indication that a replacement holder is needed.

MK
MK
nibble
Super Advisor

Re: CPU connectors/pins shredding - A5866-69001

Thanks guys. as for the question: can the cpus run if assuming those cpus were installed with damaged/missing "rubber" pins?
tkc
Esteemed Contributor

Re: CPU connectors/pins shredding - A5866-69001

when replacing the cpu, it is normally done with the cpu cleaning kit. you can check out this part, A3639-70101 : PCX-W Cleaning Kit for the N4000. it is used to clean the contact of the cpu with the system board before it's installed onto the board.

i think you are referring to a cpu chip sitting on a pvc cpu socket and the cpu socket comes with many soft buttons? the pcx-w cleaning kit for n4000 does come with this cpu socket.
Matti_Kurkela
Honored Contributor

Re: CPU connectors/pins shredding - A5866-69001

IF there is enough material left in the "pins", it might work... but the connection will probably be unreliable.

When the CPU is installed in the socket, the spherical "pins" are squeezed quite flat between the CPU and the system board. After those rubber-like spheres have been subjected to the heat of the CPU for a while, they may no longer return completely to their original shapes when the CPU is removed from the socket.

It's possible that the pins became shredded when you took the CPUs out. They tend to do that when they've been in place for a while: the bottom part of the "pin" sticks to the system board connector and the pin shreds itself when the CPU is removed.

That's exactly why the CPU socket cleaning kit exists and why it contains a replacement part for the set of "pins".

Thanks to tkc for the part number of the cleaning kit - I did not know what it's called, so I did not find it.

In theory, you're supposed to use up one of those cleaning kits each time you re-install a CPU of this type. In practice, if you're very careful and have a bit of luck, you _might_ be able to remove a CPU and still have all its "pins" in useable condition.

My recommendation:
If those CPUs were originally installed without using a proper cleaning kit (or you cannot be sure whether or not a cleaning kit was used), get at least 2 cleaning kits now. Re-install the CPUs using the cleaning kits and see if the problem persists.

If you often need to remove/re-install CPUs of this type, I recommend that you keep a cleaning kit or two at hand.

MK
MK
Andrew Rutter
Honored Contributor

Re: CPU connectors/pins shredding - A5866-69001

hi,

I have seen cpu's like yours before and this is usually caused in my opinion from the server heat generated and the force in which they are screwed down.

the tighter they are screwed in the more they will compress and flatten, and then get hot and fragile.

you can remove the plastic case carefully and replace it should you have a dead cpu or older slower one kicking about. I have replaced them in the past but not often, and you have to be extremey careful and slow when doing so, not to dislodge the pins.

I have also tried them with missing pins before and in all cases the ssyetm fails to even pass the selftests and wont start up properly. So i doubt that they were installed like that.

It is imperative that the slots are carefully cleaned and dry before you install another cpu module in it.

The servers do generate alot of heat and if a hot server room? then this could agrevate them more.

Another thing for you check though is the proc support modules? is it the same one for the cpu's thats at fault? one support module per 2 cpu's.

Andy
nibble
Super Advisor

Re: CPU connectors/pins shredding - A5866-69001

Thanks to all!