Memory issue

[BNC]

Hello - I havea DL370 G6 server that we are trying to repurpose but I am having difficulty getting it to recognize all of the memory that is installed. When I started I installed 8x 8GB DIMMS (HP # 652089-001) and the machine reported several memory errors related to DIMM #3 and DIMM #1 so I thought that the memory was in the wrong socket.

After trying several memory configurations, including one that was installe in the server before, and getting Error 207 memory configuration errors on all attempts, I went with the smallest and simplest - 2x 2GB DIMMS (HP # 500202-061) in socket 3a for each processor and powered it up. The system gave me one long (5 sec) beep followed by a second long beep and failed to go further.

The server has two Xeon E5530 processors so I know that the memory must be split between the two. It also has been configured to use Advanced ECC. Some configurations (matched on each processor) I tried include:

Config #1

1G - Empty

2D - 8GB (652089-001)

3A - 8GB (652089-001)

4H - Empty

5E - 8GB (652089-001)

6B - 8GB (652089-001)

7I - Empty

8F - Empty

9C - 8GB (652089-001)

Reports several error 207's during POST and recognizes 32GB of memory installed. The error 207's are on Processor 1 DIMM 1, 2, and 3 and Processor 2 DIMM 1, 2, and 3.

Config #2

1G - Empty

2D - 2GB (500202-061)

3A - 4GB (500203-061)

4H - Empty

5E - 2GB (500202-061)

6B - 4GB (500203-061)

7I - Empty

8F - 2GB (500202-061)

9C - 4GB 500203-061)

Reports several error 207's during post - same as config #1 and recognizes 24GB of memory.

Config #3

3A - 2GB (500202-061)

all other sockets empty

Reports on long beep (5 sec) that repeats after a few seconds and doesn't finish POST. I believe that this indicates that there is no memory installed.

Any ideas on getting this resolved?

Thanks

[gsb_2]

Try this

[PGTRI]

hi,

Try with this one on each cpu:

1G - Empty

2D - 4GB 

3A - 2GB

4H - Empty

5E - 4GB 

6B - 2GB

7I - Empty

8F - 4GB 

9C - 2GB 

and check if the dimms are clear(w/o ECC error).

Thanks

regards,

[BNC]

Good morning - thank you for the ideas. When I installed the memory as you recommended and powered the server up I recevied the usual errors. There were a number of Error 207's. When all was done the system recognized 24GB of memory even when there was 36GB installed.

The errors, in order, were:

• 207 - Invalid memory configuration - Unsupported DIMM in Processor 1 Socket 3
• 207 - Invalid memory configuration - Unsupported DIMM in Processor 1 Socket 1 (no DIMM is installed here)
• 207 - Memory initialization error on Processor 1 DIMM 3.
• 207 - Memory initialization error on Processor 1 DIMM 2.
• 207 - Memory initialization error on Processor 1 DIMM 1.

The same is repeated on Processor 2

One other thing I recall is that when I look at the last hardware scan done on this server it reported that there was no memory installed on sockets 1, 2, or 3 but only in 4, 5, 7, and 8 for each processor.

I am not certain how to ensure that there are no ECC errors short of using a dedicated testing device, which I don't have access to, but moving DIMM's from slot to slot has no effect. The same errors appear.

Thanks

[BNC]

Thanks for this suggestion but I had looked at this document. The final memory configuration was, in fact, suggested by the HP Memory Configurator.

[waaronb]

When I come across memory errors during start up (the system beeping, reporting bad modules, etc) here's the process I use, which might help you troubleshoot:

1) Reseat the modules it says are bad or missing and try again
- That lets you check if it just wasn't seated well. Works about 80% of the time for me.

2) Switch modules with different ones of the same size that it reported as good.
- This will help me identify a bad module... if the problem follows the module, I'll replace that one. This is true in about 15% of the other cases when this happens.
- If the problem stays on the same socket, then it could be a bad motherboard. I like Proliants, but sadly this does happen in about 2-3% of the cases where there's some memory issue. Replacing the system board is the only option.

3) I'll do what you did and install the bare minimum memory, with a single module on each processor.
- Sometimes memory is just incompatible with each other, and one bad module might cause issues that are hard to spot. Removing all but a couple modules can help make sure it's not a module or slot issue. This doesn't happen much, maybe another 2% of the time, and usually when I'm using some 3rd party RAM mixed with HP modules.

In your case, it sounds like one of your 3A slots is bad, since it's giving a memory error no matter what module you have plugged in. You could try removing processor 2 and only have memory in the 1st CPU's slot... it could tell you if it's the 3A slot on CPU 1 or 2, although really it doesn't matter and you'll probably end up swapping the main board anyway.

There was one case way back when on a DL380 G4 where it was definitely a bad slot, and when I looked closely I could actually see one bent pin in the slot itself. I used a very tiny screwdriver to carefully straighten it out (it wasn't badly bent, just enough that it wasn't where it should be) and after that the system worked fine. It was a brand new system and I could have (and maybe should have) RMA'd it, but I was happy with the quick fix.

If your system is out of warranty I guess you could look closely and see if there's anything obvious.

Also, just for fun, try doing an NVRAM reset in case the system had any weird memory settings beforehand like mirroring, online spare, etc. that it doesn't like now that there's only a couple modules installed.

[RbBsmn]

Yeah, try and NVRam clear, update the BIOS and try reseating the modules.
Seems like an issue with the systemboard as you get it on both processors.

[PGTRI]

hi,

Try to boot from the smart start cd to create an offline advanced survey.

Thanks

regards,