Dell OptiPlex GX270 - multiple bad cap failures

[Will Sydnor]

I am a Systems Administrator at a 3 campus community college where almost 2 years ago we purchased about 1600 GX270s. Ours are not the small form factor, however we have experienced enough failures from both the bad cap issue and the maxtor slim drive issue that Dell is replacing the drives and motherboards for all the machines. The failure rate we have had has been absurdly high. I would say before they agreed to replace them all we had already done 400+ motherboards and over 500+ hard drives.

[willawake]

eek!

welcome to the forums. can you give a better idea about how much work your team had to do because of the badcaps issue that was not covered by dell?

to what extent did your helpdesk incidents rise directly related to the badcaps problem?

just interested in some perspective on badcaps in large installations.

[tamalon55]

Forgive me if this info has been posted in another area.. This is my first post here, but I have been searching for a while..

These GX270's are really becoming a pain in my ***. After having to replace about 350 Maxtor Slimlines, the company I work for now has this bad capacitor issue to deal with (around 400 boards to replace).

The problem during our Maxtor project was that sometimes the tech would close the call even though the drive wasn't actually replaced (for various reasons, user not available, problems during ghosting, etc). However, I was able to query the registry of these computers to verify whether or not the drives had been replaced.

I need something similar for this project, some way to verify that the board was actually replaced when the call gets closed. Anyone know of a way to do this? Is there a separate board revision between the "bad" ones and the "good" ones? Some other way to tell the difference?
Where to look? Dell DMI? Registry? We also use Novadigm's Radia for software distribution if that helps, but I can't find any helpful info in there either..

Sorry for the long post.. Any help would be greatly appreciated.

[willawake]

wish i could answer your query. Personally i would be looking for a change to the product revision and maybe later bios also.

here is an interesting advisory from Dell which throws more light on the nichicon failures.
http://www.ait.iastate.edu/sales/showitem.php?id=41

Recently Dell recently discovered potential problems with a capacitor on the system board of GX270 systems with medium and small form factor motherboards. The problem is isolated to a specific capacitor supplier which was used in system boards manufactured from April 2003 to March 2004. The defective Nichon capacitors were placed on two manufacturers' GX270 motherboards between 04/01/2003 to 03/20/2004.

However we have discovered that the Nichon capacitors were not all produced outside control specifications. Rather the affected capacitors were limited to certain batches made during specific shifts during the manufacturing process. This means that not all GX270 motherboards are affected. Unfortunately, there is not specific way to determine which motherboards are affected and which are not.

Specifically the capacitors were filled with more liquid than required. After an approximate service life of 300 days and when the CPU reaches a core temperature of 64 degrees C they will begin to bulge and eventually overflow onto the motherboard causing a system crash and a "No Post" failure on boot. There is usually no data loss associated with this issue.

Dell became aware of the problem in the last several weeks based on high numbers of customer tech support calls related to the GX270 motherboard failures. Because of the extremely long time lag between system install and the beginning of the failure rate spike it was impossible to anticipate the problem we now face. In response to our engineering research regarding the suspect capacitors and motherboards, we have decided to proactively replace GX-270 motherboards at specific customer locations.

The GX270 motherboards have not been made in the past year and our on hand reserves required for normal fix on fail warranty repairs have been strained. We have arranged for these motherboards to be remanufactured however that process takes time to ramp up. Delivery of sufficient numbers of boards to move forward with proactive field replacement is not anticipated for some weeks.

[tamalon55]

Thanks for the reply.. Yep, that's pretty much what Dell is doing for us, where they've identified a certain percentage of our overall GX270 install base as "problem" systems and are mass replacing them.

I just can't always trust the technicians that they send to let me know if they can't contact our user, have problems swapping the board out so they just leave without swapping it or whatever else.

I had to use a nice little registry query I had several times while working through our Maxtor swap project. I just have to figure out if I can have the same thing to "double check" them on this project also.

But that's why I came here! I'm really glad I found this site, and hopefully one of you fine folks can help me out!

[AK0R]

How about a "hardware-based" approach, rather than software-based? Put a colored sticker, or a drop of paint or fingernail polish, on each motherboard (perhaps on a particular, non-removable component). That way, you know for sure if the motherboard has been replaced or not. It's a one-time investment of time that can pay big dividends down the road.

[willawake]

AK0R is right, it is good to label hardware. i usually label office hardware with a sticker showing the date and vendor. Then i know where to find the invoice if it needs RTM. If i am returning something i usually put a hidden marker pen dot on it so i know that i didnt get the same component back.

But in tamalon55's case i guess he is faced with too many machines and maybe different locations also to check, so he needs a non-manual solution.

[tamalon55]

That's correct.. I wish I could do a hardware solution, but the roughly 400 computers I need to replace are scattered all across the country... literally.

[AK0R]

I don't know the configuration of your machines, but if you don't shadow BIOSes, that area of the BIOS should be static. Could you use your distribution software to write a given sequence to a particular area of the BIOS of each machine? That way, you could query each machine at your convenience to know if the motherboard had been replaced (assuming that the technician didn't remove the BIOS chip from the original motherboard and install it on the replacement), since a replacement motherboard with its own BIOS presumably wouldn't have that sequence.

[willawake]

clever. looks like its gonna have to be something like that. we need modified chernobyl virus.

this thread indicates there is a possibility that the tech will not change the blank service tag of the new board in the bios to the service tag in the old board.
http://hardware.mcse.ms/message165459-1.html

[AK0R]

I thought about using the service tag, since my (newest) machine is a Dell, but there are two problems with that:
1. They are machine-specific, but not secret (something that would be known only to the network administrator, which is the point of this exercise);
2. If the tech updates the service tag on the replacement board, you can't tell if the board was replaced or not.

[bushytails]

The bios in those mentions Dell at some point, right? download the latest bios, do a s/Dell/Hell/g on it, then flash it. When the new ones come, if they say Hell instead of Dell, you know you still have the old one.

--Randy

[willawake]

ROFL

[kc8adu]

been there done that.
got a bmp that i sub for the blue dell logo on post.
its a wall of fire with the dell logo spelled hell complete with the tilted e.

[MD Willington]

OMG that is awesome, DELL =

[tamalon55]

clever. looks like its gonna have to be something like that. we need modified chernobyl virus.

this thread indicates there is a possibility that the tech will not change the blank service tag of the new board in the bios to the service tag in the old board.
http://hardware.mcse.ms/message165459-1.html

Thought about that too, but we make a BIG deal that dell updates the Service Tag when swapping sys.boards, because we use it ALOT for asset tracking. But, I may try to do something with writing a sequence to the BIOS.. That's an awesome idea. That way I can search them via the Dell Open Manage IT Assistant that we have to verify. I will look at this tomorrow while I'm at work. Keep the ideas coming though. This is awesome!

[tamalon55]

Well I figured something out. All these boards have a built in NIC right? And every NIC should theoretically have a unique MAC address right? Why not query the MAC addy's before and after?
Just thought I'd gloat a little bit.. I'm pretty happy about coming up with something.

[willawake]

pure genius

[bushytails]

Didn't know they had a NIC. I've never even heard of a gx270 before this thread. :P

--Randy

[willawake]

here come the excuses.....