Smart Array Recovery

[Louis Pickus]

I have a proliant DL380 with an Integrated SMART array controller configured with (3) 18.2 GB drives as a RAID 5 array. One of the drives failed. I replaced the drive and the machine started automatic recovery. It has been in recovery mode for a L O N G time. How long does it take to recover? Logically, none of the partitions were more than 60% full.

[Uwe Zessin]

How much is this 'L O N G time'?

I think the standard speed is the usual 15 minutes per gigabyte on an idle controller, but you can try and increase the rebuild priority. I don't think it matters if the logical disks were full or not - a storage controller only deals with blocks of data, but it does not know what a file is.

I doubt that the SmartArray has a bitmap of which ranges of a logical disks have been written, so it needs to recover the whole disk array.

[Louis Pickus]

Well, it's been going for almost 5 hours now, which means it should be done at this point (if you use the 15 min/GB rule). How do monitor progress, other than the blinking light on the drive itself?

[Louis Pickus]

It has now been close to 7 hours and there has been no change. The LED on the new drive continues to blink. The server has been idle (disconnected from the network). I do not know how else to monitor the progress, yet I see no indication that something is failing. Does anyone have any thoughts?

[Leon Rosier]

I don't know if you can shutdown the server. If so I would suggest the following:

- first of all..is the disk which you used as a replacement a fresh disk out if the HP stock?
- shutdown the server and remove the "new" disk.
- Upgrade the smartarray controller firmware and try again.

unnecessary to say you have to make sure you have a backup first!!

Leon

[Louis Pickus]

I upgraded the firmware and the rebuild completed successfully.

[Chris Bennett_9]

One other thought I had - if you have defined multiple logical partitions from the ACU or from ORCA(F8) at POST, these partitions rebuild in a serial fashion. One has to finish for the other to start. So, rebuilds have nothing to do with the amount of data on the filesystem - they are affected by the disk size and # of logical partitions on them.

Second, to view the status of a rebuild, you can launch ACU or ACU/XE and view the array status. This status page will help you determine the rebuild percentage.

For future reference only....

[Dan H.]

The same problem happen to me right now. I ran the ACU and the array was recovered but all the partitions are gone. Any idea what I can do to recover the partitions with the data on it. Thanks.

[Louis Pickus]

Again, the key to my recovery was upgrading the firmware to the latest version. After that, I rebooted the server and the rebuild process started again and then completed within a couple of hours. All partitions re-appeared and, besides some minor corruption of my Exchange databases, everything else was fully functional. I have since installed a hot spare to the array for future protection.