Hard Drive Testing

This document is currently out of date and will be rewritten in the next few weeks - Patrick 6/19/12

New Version

The actual work to be done on hard drive testing ideally needs to happen only twice a day; once for a batch of smaller drives (100GB and less) in the morning and once for a batch of larger drives (larger than 100GB) to run through the afternoon and overnight if necessary.

Setting Up

If there are finished hard drives already on the racks then proceed to Finishing a Batch.

1. Grab the red tray and head into TARDIS (the locked storage room) then open up the big brown lockbox.
2. Hopefully the drives in the box are fairly well organized by size and interface. If this is the first batch of the day you'll want to grab smaller drives, if it's the second batch you can go for the larger ones. Load up the tray with an equal number of IDE- and SATA-connected hard drives. Don't forget about the 2.5" (laptop-sized) drives on the smaller top shelf! Be sure to lock up the big brown box again when you're done.
3. Take the tray of drives over to one of the wiping racks and start connecting them to the boards. Most boards have 4 connections with various mixtures of IDE and SATA cables.
   1. Try to keep all the drives connected to a single board around the same size so we don't have 3 smaller drives finished and waiting around sucking electricity while the 1 larger drive is still finishing.
   2. Look the drives over for identifying information while you're connecting them; be sure you can clearly identify the model and serial number of the drive. You may need this information for sorting the failed drives from the passed drives later on. If you can't find the serial number on a drive then make sure you're attaching it to a board with other drives you can positively identify so you can use process of elimination when identifying the finished drives.
4. As soon as you've connected all the drives to a single board, turn it on. You can move on to hooking up another board while the first one boots up and does the initial check of the drives.
5. Once a board has finished booting Disktest has started (you will see a list of the detected hard drives and a prompt asking to start the test) proceed to Working with Disktest.

Working with Disktest

Disktest is our nifty little home-brewed hard drive testing and wiping program.

Old Version

Hard drive testing takes a while, so the actual work done in testing (described herein) pretty much only happens twice a day. The staff or build person doing this should be able to identify an IDE hard drive, figure out jumper diagrams, and remember a four-digit number for more than 30 seconds.

Quick Guide:

The following is a quick guide for those who've already read the detailed instructions before and just want to get going:

• Note the results for the completed drives
• Turn off the system and record results on labels and tallysheet
• Get drives for testing and make sure they are labeled
• Set the jumpers on the to-be-tested drives to Master or Single
• Make sure the hard drive testing computer is off
• Plug the drives in to the removable bays and slide them in all the way
• Power up the computer, make sure the tests are running correctly
• Turn off the monitor
• Put away the tested drives

Detailed Instructions

If the test is finished, it'll tell you what to write on the labels. The first will be HDA and the second will be HDC. If the test went well, it will say Enter to Power Down; hit Enter and turn off the system when it says Power Down. Record the size (in MB) on the gizmo label, and write OK so we know someone didn't just read the drive label and write the size down! If it's a bad drive, put an X through the number.

Get drives for testing, generally from the build lockup ("Tardis"). At the beginning of the day, you'll want to test drives that are nearer the low end of the range so they can finish before the end of the day. At the end of the day, test larger drives. The drives to be tested should be the same size, since you can't swap out a smaller one while the larger one is still running. We also generally try to test two drives of the same brand. Make sure the drive is labeled. There may be a sheet of gizmo labels with numbers already on them near the storage shelf, or you might need to create new ones. Try not to cover size or geometry information when putting it on the drive, and look out for airholes (they usually are marked and say not to block them).

Set the jumpers on the to-be-tested drives to Master or Single. In the case of Western Digital drives, you can just remove the jumpers or turn them sideways (to neutral position). For others, you will need to look for jumpering information on the drive labels. This may read Master, or Single, or Stand Alone, or DEV0, or something similar.

If the hard drive already has a gizmo number, note this when you put it in. The script will ask you about it.

Make sure the hard drive testing computer is off before adding or removing drives.

Plug the drives in to the removable bays and slide them in all the way; sometimes it is not totally clear if the drive sled is all the way in the bay, so give it a little push after you think it's in all the way.

Power up the computer, make sure the tests are running correctly. Each drive bay should have now its power indicator light lit. If it is not, the bay may not be locked, not seated all the way back in the drawer, or not plugged in to the hard drive's power cable. Turn off the power to the entire system before you try to correct the problem.

It will give you information about the drives it detected; make sure it finds an hda and an hdc; finding drives at other positions probably means you need to recheck the jumpers. The sizes found should be similar to what you expected. Watch it get started to make sure it doesn't have problems right away (if it does, you may want to just fail that one drive and replace it with another one).

Once you and the program agree that everything is set up to your satisfaction, it will start the tests running. You can expect to wait on the order of hours, depending on the size of the drive(s). Each drive is tested with the badblocks program, which will write data to the drives, and then confirm it did so properly by reading it. It will do this to every partition on the disk four times.

Put away the tested drives and turn off the monitor. The tested good drives will go in the Tardis; ones that tested bad should be smacked with a hammer and hurled, with prejudice and not gently at all, into the barrel set up for hard drives.