be sure to upgrade the bmc or drac fw's when upgrading the bios. That could resolve the issue.

The following is an account of my experience with PE2970 BIOS update version 4.2.1:

At the end of the day the fix is to retrograde the BIOS firmware to a previous version.

I settled on version 3.0.3 but your environmental needs may be different.

I began today with six PE 2970 chassis completely bricked after several attempts to update from different media types. The number of servers affected became so high as a result of the devious nature of the malady.  The problem only surfaces after an update has run and the server is completely shutdown. If the server is simply rebooted, the problem does not take effect. As a result I was thinking that my alternate methods of updating were succeeding when in fact , because I was restarting the machines moving onto the configuration phase of the redeployment, the failure had yet to present itself. It took a few servers to realize exactly what was happening, that’s why the number reached six affected.  I did not know that four of the servers online were prepped to fail.

OK so, earlier today  I removed all the unnecessary components ( hard drives, cd, drac, one proc and one bank of memory, cmos battery and raid battery)on one server, in order to eliminate the possible culprits preventing the POST process. I was able to reacquire the POST on one server with only one processor installed and two gigs on memory in that processors memory loop. Also noteworthy is the fact the cmos battery also needs to be removed during this re-initialization procedure in order to succeed. Subsequent to this re-initialization procedure, careful troubleshooting determined that reseating the memory with the power disconnected and drained allowed for the POST process to occur one time. Although the servers could be restarted on the fly over and over, each subsequent “shutdown” necessitated a repeat of this memory reseat routine to reacquire the POST function again. Also I noted that initially (the first time going through this process) also required the processor be reseated as well (with cmos battery removed). However during  subsequent “reloads” of the POST process, the processor did not need to be reseated, only the DIM cards.  At this point I had a method by which I could POST the server and look for boot media to continue troubleshooting.

Although I used different methods on different 2970 chassis of updating during the original attempt to prepare them for the repurposing (USB boot, Repository CD, even used the windows executable on one server that had an OS installed), the one common denominator for all the servers that were failing was the BIOS updates. So I attempted to reinstall 4.2.1 (from USB bootable media) and that did not change the status quo , still no POST without manipulating the memory modules and toggling the power supply. Next I retrograded the BIOS with version 2.02 and this cleared up the problem completely (AH HA!!!). Finally I updated the BIOS with version 3.03 and the system remained stable. I am stopping here. I have tested two servers now with all other firmwares relatively up to date in the chassis and have seen no noticeable issues. That would be the following firmwares:

BIOS                                      3.0.3

DRAC                                     1.65

BMC                                      2.50

Broadcom boot agent    4.4.4

PERC 5i                                 5.2.2-0072

Backplane                           1.05

I need to confirm some of the other firmwares to be sure they are up to date, but as of now I have recovered all six chassis at BIOS version 3.0.3 and all POST and boot as expected.

On a side note one of these 2970 chassis was a victim of this update a couple of years ago and I simply did not have time to troubleshoot it in the moment, but I placed this server in the queue of chassis to be bios retrograded and it too now has recovered from its former bricked status (two years on the shelf, thought to be good only for parts).

I would encourage your Dell team to re-examine the BIOS update version 4.2.1 as I suspect it creates a problem somewhere on the motherboard that prevents or corrupts memory management. My advice would be to remove it for testing purposes and not repost unless it can be validated. This has been a uncomfortable challenge overcoming this “bug”, and I must say being a long time Dell supported, that I am surprised that such an update did not receive more scrutiny or at least contain a warning label ( “This update may prevent your server from booting.”). I know these servers are out of warranty, but it seems to me that posting a fallible update such as this one has proven to be is just wrong on many levels.

I've had similar memory issues with my 2970 and BIOS version 4.2.1.  Mostly random crashes while running due to mysterious memory errors but the memory has been replaced several times with new tested modules and the errors seem to occur on random banks without much rhyme or reason.  I'm going to try version 3.0.3 and see if I can get the system to stable again.  Thanks for the info, I agree dell should look into this issue further to avoid a blemish on their reputation because of this firmware issue.