rx4640 - Array Rebuild - Not Working

[Jeffrey Carlson]

I have an rx4640 with two internal drives (max count). They *were* configured as a Mirror (HP terminology = "Raid 0+1").

As a measure to build our confidence in the operation of the server, and also to produce a working duplicate of the mirror [long story, another thread] we borrowed an identical (we think identical, as it has the same RPM, SIZE and SCSI identifiers) drive, pulled the second drive from the working mirror out, and put in the borrowed drive.

As expected, the server didn't go down (we did this while Win2K3 EE was running), and we expected to see the drive spin up (it did), and start rebuilding (it didn't).

Now, we're stuck. We've been into EFI and no options exist for us to force or kick off a rebuild. We've been into the GUI/HTML version of ACU XE (64-bit), and no options exist either.

What are we doing wrong?

We noticed this article: http://www1.itrc.hp.com/service/cki/patchDocDisplay.do?patchId=COL5369-IS-18302-1 that mentions the behavior required to kick off a rebuild is to hit REFRESH in the controller state field - although we do this repeatedly, there is NO indication that the systme is rebuilding the array.

I've also poured over the 64xx class user and installation guide, and nothing is obvious or specific about this behavior (or lack thereof).

We really could use some help -- anyone?

Thanks

Jeff

[Scot Bean]

Is your mirror running under LVM or VxVM?

[kris rombauts · Accepted solution]

Hi Jeff,

this is fixed by installing the rescan service for the SmartArray controller:

http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareDescription.jsp?locale=en_US〈=English&pnameOID=346361&prodSeriesId=88837&prodTypeId=15351&basePartNum=COL9325&locBasepartNum=is-25222-2&os=Microsoft+Windows+Server+2003+64-Bit&tech=Utility++Smartarray

It will rescan every 5 minutes and start a disk rebuild automatically, so no reboot is needed anymore.



Other solution is to upgrade the SA64xx firmware to 2.26 and use the last ACU Version 7.15.18.0 utility and use the "rescan controllers" button in the ACU to force a rescan of the SCSI bus.


HTH

Kris

[Jeffrey Carlson]

Folks/Kris:

Thanks for the quick reply. We located and installed the patch, and didn't appear to have to reboot. Although I will reboot the node after posting this.

However -- it would seem that either I'm not quite seeing the option, or that something else has happened entirely. Specifically, when we originally removed the drive and inserted the spare (to force the rebuild), we noted that the XE ACU had named the array as RAID 0 (striping without parity).

This would conflict with what we had originally installed the systems as, being RAID 0+1 (mirroring).

Bizarre as it seems, after installing the patch mentioned previously in this thread, I do not see any noticeable change in the behavior of the XE ACU nor do I see the array marked as a RAID 0+1 "rebuilding".

Possibly what may have happened is that by removing the disk, the array was somehow converted to a non-redundant level of RAID and now is stuck in that mode?

Migrate options do not list RAID 0+1 as an option. Perhaps I am selecting the options wrong?

Again, thanks for the help. 64-bit is uncharted territory in my administration experience, and I long for the days of the old 32-bit based GUI. Lol.


Jeff

[Jeffrey Carlson]

I've taken a screen shot of what I'm seeing. It's attached as a JPEG.

Thanks!

Jeff