Problem with ProLiant 5000 and HDD's / SA 2/P

[Max Habermehl]

Hello,
a customer is experiencing problems with his old PL5000, RAID5, 5*HDD's, SA 2/P.
In the past one drive often quit working and showed amber light.
We replaced the drive with a spare drive from another server (not in use any more), a rebuild was performed and the drive ran for a while.
Then again it was indicated as faulty (amber light).
We talked to an HP-tech and came to the conclusion to first try to exchange the backplane as it was the same drive on the same slot which was going faulty and faulty again and replacement of the drive had not shown any "long-term" effect.
If this wouldn't work we thought of trying to replace the SA 2/P controller.

And here is what happened today (just a few days before we could try out the backplane-thing):
The customer is reporting by phone that again that specific drive is showing an amber light and that he has tried to pull it out and put it back in again.
Then (this is a new behaviour) the system reports at restart (after the "Press F1 to recover / rebuild" - message which the customer confirmed by pressing "F1"): INACCESSIBLE_BOOT_DEVICE.

This does not sound optimal at all. ;)

I know this error often is related to problems with boot.ini pointing to the wrong drive or maybe with faulty controllers.

There is my question:
The controller does cache, what drives are placed where in the backplane / array, right?
The drives' do cache the information (if intact RAID5, which means no more than 1 drive is faulty), on which controller they had been attached (and SCSI-ID and so on), right?

So what will happen if I try to move those drives to a replacement-controller (the customer has a second PL5000 at hands for such purposes )?
Will this work or will we get in trouble as the "new" controller and the drives "do not know each other" and therefore the logical drive will not be recreated?

Any suggestions / help would be highly appreciated.

Thank you,
Max

[amhakassa · Accepted solution]

Hi Max,

Yes you can move the drives to the other Proliant 5000 with the same controller which is not a problem. The reason is, the raid information is stored in each hard drive's RIS (raid Information Sector table) so the controller does not save any RAID information

But, I would suggest to try this

1. Power down the same server and take the drive that was showing umber out then reboot server
2. You probabily get a POST message to notify you that one drive is missing and that is fine continue and see if you can get the OS

Let me know if this works or not and we can do other things before you move all the drives to the other server

If this works make sure if there is any bent pin on the scsi connectors on both the controller and the backplan side.
I would suggest to reseat the controller and the cables as well

Regards
Amha Kassa

[Andrzej Kowalik]

Hi Max,
You can replace controller without any problem. From 2/P to 3200 they use the same system driver (Windows).
Drives not hold informations to each controller they was connected. They hold data about array/logical disks configurations (and therefore you can safely replace controller).
But if I remember this controller not support disks order changing - you must place disks in the same way.
regards,
Andrzej

[Max Habermehl]

Hello,
thank you very much for your replies.

In the meantime, I am at the customers site (it is 9 pm here - my girlfriend will possibly kill me for that).
The exact error code is "STOP! 0x0000007B (0x82AC7030,0x00000032,... )INACCESSIBLE_BOOT_DEVICE"

This error unfortunately apperas during the windows boot-up (not as I expected right from the start). The system is saying "starting windows..." then the Windows Startup-Logo shows up and the this error shows up as a BSOD.
I am afraid that (as a consequece of the customer not wanting to use a UPS) maybe the registry / ads is faulty. Can anybody confirm this?

What I tried:
Took out the faulty drive, restarted the server. Error 1789 appears "Check cables or replace the following SCSI drive(s)...
Press F1 to continue - drive array will be disabled.
Press F2 to fail drive(s) thar are not responding - Interim recovery mode will be enabled if configured for fault tolerance".
I hit the "F2" button (it may be that the customer hit "F1" this afternoon :eek:).
Message 1787 points out, that interim recovery mode is enabled and that the following drive (...) should be replaced.
I hit "F1" to continue ==> the same error shows up (BSOD).

Right now I am trying to use my winternals erd commander cd to check if there is still any data on the array.

Any further suggestions are highly appreciated.

Thank you,
Max

[Andrzej Kowalik]

Did you replaced array controller?

[Andrzej Kowalik]

I mean - you wrote before, that you want to move disks to another server, this is new server? Or problems occured on old hardware?

[Max Habermehl]

Hi,
I did not replace the controller so far.
Regards,
Max

[Andrzej Kowalik]

http://msdn.microsoft.com/library/default.asp?url=/library/en-us/ddtools/hh/ddtools/BCCodes_af56ca9a-4216-45b2-ba40-027da48a961a.xml.asp
good luck

[Max Habermehl]

@Andrzej

Hi,
sorry, but our yesterdays messages seemed to have overlapped. I had not seen the post before your last post, before I replied.

We have 2 servers here - the one that ran for some years and another one that is not running (only used to take spare parts out for the running server).

What I will do is, according to your recommendations, replace the controller of the gone-faulty-machine with the one of the spare-part-machine.

I will report back then.

Thank you,
Max

[Max Habermehl]

Hi again,
it was not the controller which causes the problems. I put another controller in the machine: Same error.

Tried to boot in "Safe mode": Same error.

"Last-Know-Good": Same error.

Bootet with SBS-Installation-CD.
Recovery-Console --> DIR: No data to show.
Then chkdsk --> Nearly all data could be recovered, system was startable again from that point.

So in the meantime I had the opportunity to access the server via network and to backup all the data necessary.

I assume that after one drive had failed there was another drive near to failing and that may have caused data corruption.
Am I correct?

Anyway: Thank you again for your help - points will be submitted. ;)

Best Regards,
Max