Buffalo TeraStation 1.0TB Faulty Drive Failure on RAID5

14:49.09 - Wednesday 9th May 2007   (Link to This Entry)

Recently, one of the drives in our Buffalo TeraStation NAS (Network Attached Storage) decided to spit its dummy and throw up a Diagnostic fault. The little red light would be blinking away while the TeraStation itself did nothing, save whir quietly in the corner. There was a brief moment of “Oh my god – our DATA!” before I realised the thing was still in warranty – Buffalo offer 2 years on the TeraStation and ours was still a few months inside that.

After an emailed conversation with Gerard at Buffalo Technologies in Ireland, it was decided we would remove the drive and reformat it (using NTFS) so see if it was a simple corruption issue rather than a physical fault, and if the drive did turn out to be dead, they’d ship us a nice new one. Thankfully we had formatted the unit using RAID5, which allows one drive to fail without loss of data, and so we fully expected to be able to recover everything afterwards.

The first job was to get into the TeraStation in the first place. Removing the three screws from the rear of the cover and the four that held the feet onto the base allowed us to slide the cover backwards and off, only to find very little actually showing. Another four (smaller) screws allowed us to remove the front plate, and after unplugging it, another five screws came off to remove the metal plate holding the drive cables in.





So far so good – we can now see the connector end of the drives lined up, with drive 1 at on the right and drive 4 on the left. Since drive three was the duff one, I disconnected both Data and Power leads and reconnected the front plate. Plug the unit in, press the power button (inside the unit) and – hey presto – after a few seconds of flashing lights, up pops the TeraStation onto the network.



Because we opted for RAID5, we are able to access the only project we don’t have a copy of (because at 80GB it’s really too big to stay on anyone’s hard drive if they want to do any work) and copy it across for safe keeping – everything else we already have.

RAID5 spreads the files over all drives with distributed parity checking, so if one drive fails, you can still get your data. It works on an N+1 basis, so with four 250GB drives we have 750GB of space before formatting overheads, plus the 1-drive safety net. One drive is a small price to pay for the protection RAID5 offers, since all we need do is replace the faulty drive, and the TeraStation will go fill in the blanks for us.

Rather than dismantle the TeraStation completely, I decided to remove the side cover from my machine, plonk the two side by side and run an IDE cable from the PC in order to mount and format the dead drive. Power was a little more tricky, so I opted to leave it connected to the TeraStation PSU and booted that after removing data from the other disks. The new drive was detected by Windows and after removing all partitions, creating one great big one and reformatting using NTFS, I reconnected into the TeraStation array for testing.



One thing of note: Buffalo told me that any 250GB drive will do as a replacement – you don’t have to find the exact make, model and revision of drive in use by your unit, which I’m sure you used to have to do with SCSI RAID once upon a time. Presumably you’ll need something with the same head and sector layout, but that shouldn’t be too hard to find. Our TeraStation uses Western Digital Caviar drives with the code WD2500BB-00GUC0, which I was unable to find for sale, but eBuyer sells a WD2500BB variation for about forty quid, so out-of-warranty issues are pretty much covered from now on.

Once the drive is reformatted, reconnect it back into the TeraStation array as before and fire it up. Hopefully you will see the unit boot with the bad drive still showing a red Status light, but the TeraStation itself will appear on the network and you will be able to access the files using the three good disks as before.

Next up, log into the TeraStation control panel using its built-in web interface and proceed to ‘Raid Configuration’. You will see a note about the error in RAID #1 which you can click, taking you to a list of drives under the heading “Disk Structure” of which only one – your bad one – can be selected. Tick it, click ‘Restructure RAID Array’ and go have lunch while the TeraStation fills the reformatted drive.



During this process, all eight lights in the Disk Information circle should blink green, while the DIAG light should blink red. If any of your eight Info lights are red, you’ve got a physically duff drive and will need to replace it.

Once the data is spread back over the four drives and you have green lights across the board, power the unit down and reassemble it. Your TeraStation should now be back to normal, but keep an eye on that drive, perhaps making a note about which one has failed and when, to see if it happens again. If the same drive fails again, it may be worth replacing it after all.