Disk Arrays
cancel
Showing results for 
Search instead for 
Did you mean: 

Another 12H question

SOLVED
Go to solution
Ron Irving
Trusted Contributor

Another 12H question

Does anyone know how long a read/write/verify test, (destructive,) takes on a 36GB drive in a 12H? Just curious.

Thanx

Ron Irving
Should have been an astronaut.
8 REPLIES

Re: Another 12H question

Ron,

And how are you going to run this verify on a 36gb drive on the Autoraid? Remember it has to be assigned to a LUN in order to have a device file to run the verify on. You can't single out a drive in the Autoraid and just run a test on it. Just think of the Autoraid as a big "disk cloud". All of your data is distributed amoung all the drives. When you unbind a LUN it will find the data from that LUN that resides on all the drives and remove it. That's why it will "christmas tree" when you add, remove, or it's balancing. It manages where the data is, not you. The Autoraid itself will run disk verifies in the background when it's not being used. It basically sniffs for drive errors in times of low activity. Is there a specific reason that you need to verify this one drive? If you try to use STM to do these verifies it won't see the drive.

Re: Another 12H question

As a side note, your hot spare isn't even a traditional hot spare. It is a defined unused space striped across all the disks in the array, not a physical drive. Hot space space is used by the Autoraid as swap space if it's not in use due to a failed drive. Your hotspare space should be large enough to cover your largest drive. If it's not your hot spare rebuild will fail. Second side note, if you do end up with a failed drive you want to let the Autoraid completely rebuild the data on the hot spare before inserting the replacement drive to minimize any possiblity of causing the array problems. Any rebuild should be allowed to complete in full before doing a removal and or replacement. Trust me on this. Depending on how hard the array is being used, rebuilds can take up to 48hrs to complete for the larger drives. You can only create up to 8 LUNs on a Model 12H so if you continue to add larger drives to it you will hit the 8 LUN threshold and probably be forced to unbind LUNs so you can create larger LUNs. This will entail doing a data restore for the LUNs that are unbound.
Bill McNAMARA_1
Honored Contributor

Re: Another 12H question

This is possible after downing a disk via the Array CLI... or even via STM after having downed a disk for destructive testing.

I would guess at 8hours, although I have never done it myself.

Perhaps you could be the first to tell the world!

I'll have a look around myself..

Later,
Bill
It works for me (tm)
Ron Irving
Trusted Contributor

Re: Another 12H question

Thanks for the responses. However, I'll try to be a little clearer. This array is setup on a test machine, (D380,) with both controllers installed and setup. We received 8 36GB drives earlier this week, and wish to run the read/write/verify test on the drives as it is explained in the 12H manual. The command is:

# arraycfg -D a2 -d -Z 00000026412A
-to down the particular disk, and

# drivetest -D a2 -w 100 00000026412A
-to perform the test

I started the test on this particular drive yesterday at 1PM EST. Almost 20 hours later, it's still chugging away. At this point, we're giving it until it hits the 24 hour mark, and then stopping it to try something else. Perhaps a 'mediainit' command? Any suggestions?

Thanx again,

Ron Irving
Should have been an astronaut.
Ron Irving
Trusted Contributor

Re: Another 12H question

Another point to add is, the reason we're doing this is because HP didn't have anymore 36GB drives for the 12H to sell us, we had to go through a broker to get them. None of them are HP certified, and we wish to test them before we put them in our production server. We have a total of 8 of them to test. Seeing as we'd like to install them before Christmas, perhaps the route we're going wouldn't be the most 'time saving'. Any alternative ideas would be most welcome. Thank you.

Ron, (shoud've been a janitor,) Irving
Should have been an astronaut.
Solution

Re: Another 12H question

If the Autoraid is running the latest firmware (HP62) and all the drive firmware levels are up to snuff, the Autoraid should be a pretty good judge of whether or not the drive are good or bad. Unfortunatly there isn't another method that I know of to complete the task you want to acomplish. Is this array going to be put under HP contract eventually?
A. Clay Stephenson
Acclaimed Contributor

Re: Another 12H question

Hi Ron,

If I were doing this, I would simply run an arraymgr -g start and let the array begin doing its normal integrity checks. You will
have rather good data in at most a few hours as the the health of all disks and array components. I admit thats it's not as definitive as a destructive I/O test but since the 12H is designed to handle failures; it's also not nearly as vital.

If you are not planning to put your array under maintenance, I would go ahead and purchase a couple of extra drive modules to serve as spares.

If it ain't broke, I can fix that.
Ron Irving
Trusted Contributor

Re: Another 12H question

Actually, this array is not going under maintenance at this time. I work for the support company here on-site, and we're just bringing this one up to speed to use as a spare if needed. Thank you for all of your responses to my query. All of the info provided was, as always, much help to my tired old brain.

Ron Irving
Should have been an astronaut.