3200 Controller Failures

[Robert Garito]

We currently have 5 brand-new 3200 array controllers that have failed with the same basic symptoms. (Our test environment for these controllers is either a 3000 or 6500 server--the symptoms are the same in both).

1) When the server is first turned on, the diagnostic LEDs on the controller card begin to cycle through the card's self-test and then halt. The LEDS halt, with one of the two following patterns:

*=Lit O=Not Lit
<<<< CR1 CR16>>>>
*O**OOOO OOOOOOOO
*O***OOO OOOOOOOO
****OOOO OOOOOOOO

Is there documentation available that tells what the LEDs mean both during the self-test and once the card is running? This information would be VERY useful, as we have over 150 Compaq servers on site and knowing what the LEDs mean could be useful in troubleshooting...

2) When the system boots, the BIOS on the card reports:
1783-Array Controller Failure
(Command Failure cmd=11h err=20h)

It's strange that we have 5(!) controllers with exactly the same problem. All were brand-new spares (never used). Unfortunately, they're out of warranty. FYI--we do have some working controllers tested in the same hardware, so I know it's not the server with the problem.

Thanx in advance!

[Ting Lee Yiaw]

I have 2 units of 3200 that had the same problem. I beg a used card from Compaq in order to bring my server back online. Both units failed in duration of 2 months. What a coincident!

[Anton Reiman]

This does seem to be a problem with the Proliant 6500 and the SMART 3200. As I have a couple of these servers with the same problem. I have replaced the controller in one of the 6500 servers at lease two times.

It only occurs when you reboot the server. I have not yet had an "online" failure.

[Steven Meyer]

I have also seen 3200 failures on reboot.

Had a production and backup server fail at the same time because units were rebooted. Two new 3200 controllers and 24 hours later we were back on line.

Any idea on what is causing the problems? We have a large install base of Proliant servers with 3200 controllers.

Steve Meyer
Field Engineer
GE Medical Systems Information Technologies

[Leon Rosier]

Hi,

Since these 5 cards have been in stock for a long time. The cachebattery might be totaly empty and not able to charge anymore.
If you have a working card somewhere. Try to place the cache on one of your dead controllers. If this does not work, i think all controllers need to be replaced.

Leon

[Robert Garito]

Unfortunately, I tried to swap the cache module with a known good working module and still had the failure. (I know the batteries are fine on the one I tested with).

Of course, if HP or Compaq or whomever they call themselves would provide some information on what the LEDs mean, we could probably troubleshoot this better....

[Leon Rosier]

I still think you suffer from some defective boards (don't ask me why). I also think that if you mention this to HP/Compaq they will help you with spare boards (not sure though).

Last thing you could try: Disconnect all scsi cables (drives) attached to the smart array controller. Perform a system erase, if possible with sa3200 installed. Also clear nvram on system board.
And last but not least: say a little prayer (i'm not religiuos, but if it doesn't help it doesn't hurt)

Good luck

Leon

[Victor Pendlebury]

Be sure to use latest Firmware on SA3200 ...

[David_452]

Beware the BIOS update, I tried the latest update in a 6500 last weekend, and it failed, to the extent the card was not recognized on the next boot. The card had been in place for at least 2 years and the bios had not been updated. Any thoughts on how to correct this dead card would be appreciated.

[Adrian Hall_2]

I've just posted a related question:

I have just acquired another SA 3200 to replace the on e giving the "Command Failure cmd=11h err=20h) " error, but this new one is giving the same.

Anyone know what other bits of the system could cause this error. I fail to accept that the cards could both be faulty in the same way - and I would never buy[1] another compaq/hp product if this poor shap has 5 failures in the same batch.




[1] Granted I didn't buy this server, but hey ho.