RAM no Longer Detected on BL460c Generation 8

[drosich]

Hello everyone,

I am new to servers and enterprise-grade equipment and encountered an issue a few days ago that I can't seem to solve. I have a c7000 chassis fitted with two gen 8 blades and five gen 5 blades. On Sunday I decided to move the whole server setup to another part of the house, and to do that I realized that I had to disassemble the whole server so that it would be light enough to carry and move to another rack. Upon reinstalling at the new location, first all of the server blades were labeled by the chassis as "power delayed," and I needed to clear the VC profile. After doing this, all of the blades turned on just fine except one of the generation eights. After connecting a monitor and keyboard to the blade I saw that it would boot through the bios and begin the HP Proliant system check over (power, thermals, processors, and of course RAM). In my current configuration, I have 128GB of DDR3 1066MHz RAM in that blade. However, while the blade is booting up it shows that it only registers 96 GB of RAM in the server. I will also leave a link to the error message that it shows for the undetected RAM.

https://photos.google.com/share/AF1QipNdiFiyvrknksfArKAWq4LOJRAAG8WOMzRpUD2RPrlU24ckBRGs4qVvgmLOc6r7sA?key=Nmp1aXBIa1VBUjRYZHBWUUNkRHBqZlRta2c5amln

What I have already tried: As you can see in the photo, processor two slots 5-8 are the only ones that register as faulty. To try to narrow down the issue I have verified that the ram installed is working, tried adjusting what slots are populated, exchanging the processors, booting with dip-switch 6 in on and then restarting, and even put in new RAM, all to no effect. The strangest thing is that even without any RAM installed in the slot, it still gives the error message upon bootup that the installed RAM in slots 5-8 is incompatible. I originally thought that maybe some pins on the CPU were bent, but I don't know if that explains it giving the error message even when there is no RAM installed

Any help solving this would be greatly appreciated. Thank you all in advance! 

[DANDKS]

If the server with all the Memory modules installed were working fine before moving the Enclosure, it is possible that the slots are having issue or the CPU itself.

Did you disassemble the entire server components, did you remove the CPU's & the Memory modules from the server while moving it to another location?

Have you tested the same Memory modules in other slots & are they working fine?

Are both the CPU's working fine when they were swapped between the sockets?

Closely watch the CPU pins for any damage like bent pins

Clean the Memory slots using a brush so that there are no dust particles in the slots

Clean the pins on the Memory modules using an alchohol based solution or even an eraser

Update the status

Thank you

[drosich]

When I moved the enclosure to a new location I took four 16 GB memory modules out of each of the gen 8 servers to use in a computer I am building. This left 8 memory modules in each of the two blades. Out of my two generation eight blades, the  gen 8 server that is working has the exact ram in the same slots at the troublesome one and manages to work without any issues. This morning I removed the faulty generation eight server and took out CPU 2 for inspection. The socket was perfectly clean and none of the pins seemed bent. I also inspected RAM slots 5-8 for processor two and none of them seemed dirty or dusty at all either. The ram installed is genuine HP blade server RAM and works without any issues in my other blades. The most confusing part for me is that in this faulty blade I have eight total sticks of 16 GB RAM installed, which means that only half of the slots are populated (in my current configuration this means that the RAM sticks are only installed in the white slots). However, when the server is booting up and displays the message that the RAM is not supported, in addition to giving the error for Processor 2 slots six and eight,  it also says that the RAM installed in slots five and seven is also faulty and not recognized, even though those slots are empty in my current configuration. The server only recognizes 96 of the 128 GB of RAM, which means that all the RAM in processor 1 is detected and only half of the RAM in processor two. 

[drosich]

I took out CPU two and put all of the RAM in CPU one. What happened then was all 128 GB of RAM was detected, showing that the RAM is in fact compatible with the server, and no stick is damaged. I then proceeded to put CPU 2 back in the server without any RAM installed and got the same result: all 128GB was still being detected as it was populating CPU one's RAM slots and the server was detecting at CPU two was not compatible with the RAM installed in DIMMs 5-8, even though they were empty. I am still baffled and would really appreciate some help on how to solve this. Is there any way that this could be a stored error code or something that I might have inadvertently changed when relocating the server? Once again, I don't think that any of the pins of the CPU socket are bent but from the way that the server has been acting it would not surprise me... Someone please help

[DANDKS]

Hi,

There are three possibilities in this Blade.

1. The Memory Slots are faulty

2. The CPU 2 is faulty

3. A faulty BIOS

To isolate further, swap the Processors between the sockets. CPU in socket 1 is working fine as per your confirmation. Remove the CPU 1 & keep it aside.

Remove the CPU 2, install it in Processor socket 1 & install all the Memory Modules for CPU socket 1

Turn on the server to check if the CPU & all the Memory modules are detected (2nd CPU in socket 1)

If all the memory modules are detected & working fine without throwing any errors while booting, turn off the server & revert to original configuration & test the server again.

If the issue is observed when CPU 2 is installed in CPU socket 1, then the CPU is faulty & requires replacement

If the issue remains with the original configuration, it could be a faulty system board

Try clearing the NVRAM, remove the CMOS battery on the system board & turn on the server once. Once the POST is complete, turn OFF the system & install the battery. This process will clear the NVRAM (will clear the BIOS settings). Turn on the server & verify the status.

If the issue remains, this requires a hardware fault analysis to be done. Please log a support ticket with HPE server support team for further assistance.

Thank you

[drosich]

I just moved processor two to processor slot two along with all of the RAM and it showed up as having 128GB of RAM as it should. However, upon installing former processor one into the slot for processor two and moving half of the RAM over, I got the same error message, unfortunately. I then proceeded to remove the CMOS battery and start the server, and once it booted to turn it off and reinstall the battery and turn it on, got the same error. Went into setup and returned all of the settings to factory defaults, same error. I am thinking that it is faulty memory slots. 

I went ahead and ordered a new system board for the blade and hope that this will fix the issues as it is not due to the CPU, RAM, or even the BIOs apparently. The part that I am most worried about is that this happened spontaneously and not due to a clear incident like dropping it or a bad BIOs update: all I did was remove some RAM and it broke the system board somehow.

[DANDKS]

Hi,

The server can be moved without removing internal parts, which is also safe.

Let us know if you have installed the new system board & everything is working fine.

Thank you