Proliant 1600 troubles

[Nick Davison]

We have 2 Proliant 1600 450Mhz servers which have been happily running Netware (and have remain untouched) for at least 4 years. Unfortunately one of them "halted" due to too many abends (abnormal ends) which came out of the blue ! I took this server out of use and started running MemTest86 on it to see if there was a memory problem. The test ran sucessfully for over 9 hours before stopping with an "unexpected interrupt", presumably not related to the memory. I swapped out the DIMMS and it still happened. The server will run for anything between 1 and 9 hours before falling over (could be a thermal thing, but what ?). Any advice would be gratefully received as to what path I should follow to track down the problem.

[Tarek Kawach]

Hi Nick:


If you have never replaced the processor cage; it is time to replace it; there is a bad Fairchild Chipset on that board.

Processor Board, Dual SCSI, with Cage and Fan spare # 313622-001

Replacing that board will fix your issues.

If it is at all possible to try a different board from other server that you are not using to test.

Tarek.

[Nick Davison]

Tarek,

Thanks for your advice. The other 1600 we have is the main server for our business, so I hope that it doesn't go the same way as well. From a quick search on the net, it would appear that the price of the board is around the price of a second hand server on Ebay, but any used machine may have the same problem, so it may be time to gamble and replace that board with a new one. I guess you've seen these symptoms before to be able to pinpoint the fault so precisely. Is this how the Fairchild problem manifests itself, and what exactly does this chipset do ?

Many thanks again. Regards, Nick

[Tarek Kawach]

Fairchild is the Chipset on that board; and everytime replacing that board fixed issue correctly:

you could be getting these errors.

Event ID 4199.

NMI errors.

Unrecoverable memory errors.

Replacing all memory and cpu on that board dosenot fix issue; then once the processor cage is replaced; problem resolved.

I guess it is a bad chipset on that board.

Fairchild Semiconductor (part number FST16212).

now if you have the Texas Instruments Incorporated (part number 74CBT212) chipset on that board you should be okay; you will have to read the chipset on the board to find out which one you have.


hope this helps.

Tarek.

[Tarek Kawach]

Nick:

Before you invest any money on replacing the processor cage; make sure that you have tried with minimum config; no 3rd party hardware, one memory stick and one processor and swapped the processors. to eliminate cpu cache or memory as being the issue.

Although I am sure that prcoessor cage will resolve your problem but that is still a guess on my part. since I don't know the full history of the server; and what other hardware you have on it.

Best regards,

Tarek.

[Nick Davison]

Tarek,

The server has no other hardware installed other than that which was supplied, so it's a fairly "clean" system. I have examined the processor board, and can't find a Fairchild chip, nor a Texas chip with that part number. Where should I be looking to locate the chipset in question. I have identified the following "larger" chips with various markings :-

LSI Symbios 53C876
Digital 21150-AB
Intel FW824438X
Texas TLAN TNETE100AGPE
Cirrus Logic CL-GD5446-HC-B
??? ES9820AV

Image of board attached.

Regards, Nick

[Terry Hutchings]

Look on the side of processor slot #2 and you'll see the chips there. I'm not able to see whether it's fairchild or TI, but it will be one of the two.

[Nick Davison]

Terry,

Thanks for your help with this. Forgive me for being a little "blind" here, but is it possible for you (or Tarek) to refer to the image of the processor board I attached to my previous missive, and point out exactly where the chipset in question is (maybe modify my image with a big arrow ;-). The chips in the vacinity of the second processor slot have neither Fairchild nor TI markings on them as far as I can see.

Regards, Nick

[Tarek Kawach]

Nick This is Texas Instruments chipset from the image you have sent .