DL 585 G2 Memory Issues

[Borys Dziegielewski]

Hi Everyone,

We are experiencing some problems with the new DL 585 g2 we ordered. The problem exists with failed memory . In the last two months we had three failed Ram chips on different servers. There is no OS installed on them yet as we are running tests for our new ESX farm. The servers are basically just connected to power. The memory banks are populated with 2 (2GB chips) and 6 (4 GB chips) on each of the four processors. When one of the memory slots fail the entire server shuts down with an internal health light displayed on the outside console of the server. The inside status panel lights up with all 8 memory slots indicating failure for all 8 under that given processor.

We are then forced to take out one ram chip at time and swap it with one from another server as we do not have any spares yet to figure out which one has failed. We then power the server back on and see if we found the right one. This is really serious as we can see that the whole server hosting VM's will crash when a failed memory chip is detected.

Our DL380's all just indicate the failed Ram ans still keep running.

All the RAM is HP brand.

Any help or direction would be greatly appreciated.

[Blazhev_1]

Hi,

use Firmware CD to update the Firmware :

http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareDescription.jsp?lang=en&cc=us&prodTypeId=15351&prodSeriesId=3219717&prodNameId=3288124&swEnvOID=1005&swLang=8&mode=2&taskId=135&swItem=MTX-2c5987b8362348dbb3e80d277b

The DL380 maybe indicates not a real failure - they indicate a memory error threshold exceededm, which is different.

In case that DIMMs continue to fail, you can enable some kind of advanced memory protection from RBSU to ensure that server will continue to run in case of 1 chip failure...

Regards

[Esteban. ISS]

Correct. On that particular box you can configure "an array" of memory dimms as well.

ECC must be enable too. The cool thing about the server you just get is that you can have redundancy on the memory dimms as well by configuring it on the RBSU.

For more information and steps about "how to" refer the server user guide on page 40

http://h20000.www2.hp.com/bizsupport/TechSupport/CoreRedirect.jsp?redirectReason=DocIndexPDF&prodSeriesId=3219717&targetPage=http%3A%2F%2Fbizsupport.austin.hp.com%2Fbc%2Fdocs%2Fsupport%2FSupportManual%2Fc00778940%2Fc00778940.pdf

---
Steve

[Borys Dziegielewski]

Hi Everyone,

Just thought I come back and share my findings. All the bad RAM from the servers was replaced. It turns out that one of the problems we dealt with was regarding the RAM manufacturers. HP states that RAM in parallel banks must have the same part number for the server to work. The DL 585's we received had RAM from four different manufacturers. Each server had a mix of these four. Even though all the RAM was certified by HP and had the same part number, a problem still existed. If the parallel memory banks are not filled with the same manufacturer, the server will not boot up. This means that if one bank has Q logic and the next parallel memory bank has Samsung, the server will crash.