Thursday, September 11, 2014

Computer brain surgery or How to performa a RAIDectomy

I have an old 2010 Mac Pro in my home office that is my main photo editing machine (using Adobe Photoshop CC and Lightroom 5). It also serves as a remote desktop terminal to my office PC. It has 2 Intel Xeon CPUs with 6 cores each running at 2.93 GHz and 32GB of RAM, so even by today's standards, 4 years later, it is quite a powerful machine.

At least in theory it should be. Back when I bought the machine I thought it would be a good idea to get the Apple RAID card and 4 drives with 2TB each and set them up in a RAID 5 array for data protection, giving me a total usable 5TB of disk space, which I  set up as a 2TB boot drive and a 3TB drive for data.

Apple RAID Card and Drives
Apple RAID Card and Drives

That RAID card, however has been giving me nothing but trouble in those four years, and it got so bad this summer that it was time for a radical move: RAIDectomy!

The issues that I experienced with the RAID card were the following:

  • Every so often the RAID card would complain that the on-board battery was not fully charged, and would disable the write-cache, resulting in a severe performance hit that slowed down the entire machine to a crawl.
  • Every couple of months the RAID card would enter a mode called "Battery reconditioning", where it also disabled the write-cache for a day just for good fun, and there was no way to stop or postpone that process. If you wanted to get any work done that day, you were out of luck.
  • About every 3-4 months, the RAID card decided that it was time to rebuild the RAID, so it went into a 48 hour mode of scratching all the disks 100% and the computer was unusable during that time.
  • Then the RAID card informed me that the battery was dead and disabled the write-cache permanently last year.
  • Even after replacing the RAID card battery back then, these issues did not go away, but were rather just suppressed for 4-5 months, before they resurfaced.

And now the RAID battery has begun a few weeks ago to give me the impression that it was going to die again soon, so I decided to take a more radical approach this time and get rid of the RAID card once and for all.

I also decided to replace the hard drives and get something with a bit more speed and less wear and tear to reduce the risk of data loss, so I bought a Crucial 1TB SSD drive and a new Seagate 3TB 7200rpm hard drive both with SATA interfaces. I figured I would use the SSD as my boot and application disk, and the larger 3TB hard drive for photo storage. To ensure that the smaller 2.5" SSD drive would properly fit in the 3.5" bay, I also purchased a conversion bracket.

So the process I had in mind was to copy the data over from my two old RAID logical drives to the two new drives and then remove the four old drives and the RAID card.

Easier said than done...

Obviously the first step was to do an extra backup (in addition to the TimeMachine/TimeCapsule network backup that was always running). So I connected my 4TB external USB 3.0 hard drive and started the copying process - only to realize that Apple in their infinite wisdom only equipped the Mac Pro with USB 2.0 ports and the only fast external port on that machine was a FireWire 800 connection. Of course, nobody still uses FireWire on this planet and no external hard drive in my collection supported it, so I had to wait 9½ hours for 3 TB of data to copy over to that external drive using USB 2.0.

This also meant, that my original plan to restore the data onto the new disks from the external drive was going to be more painful than I was willing to entertain.

So I decided to do a 2-step approach. The Mac Pro luckily has two 5.25" bays for optical drives and I had only one of them filled, so I had a SATA connector available on the inside of the chassis that was not controlled by the RAID card. This allowed me to connect the new 1TB SSD inside of that bay and just let it sit there without any screws attached. Upon powering up the machine, I used Apple's Disk Utility to format and partition the new drive, and then planned to use the "Restore" function inside that software to clone all the data and the recovery partition from my old boot volume on the RAID onto the new SSD boot disk. Indeed, there were many online support discussions that I found that praised the ease of using the Apple Disk Utility for that process.

What all these discussions and help forum posts failed to mention is that you cannot use that process to clone the active startup disk on a Mac. There is simply no way to do it with the built-in tools.

After a bit more research I found a nifty utility called Carbon Copy Cloner that promised to do exactly what I needed, and it offered a free 30-day trial, so I downloaded it and gave it a spin. Indeed, it was not only able to properly copy my entire boot volume from the RAID to the SSD, it also correctly copied and built the recovery partition for MacOS X. Huge tip of the hat to this software, and after I saw it working so flawlessly, I did, of course, purchase a license.

After the boot disk was cloned, the next step was to repeat the same process with the data drive: I disconnected the SSD, put the new 3TB drive into the 5.25" bay, used Carbon Copy Cloner to copy all the data over, and then removed it again.

Now it was time to perform the actual RAIDectomy and remove the 4 original drives and extract the RAID card. That process went very smoothly, and I was also able to quickly mount the 2 new drives in the main drive brackets and insert them so they connected directly with the backplane.

I was pleased to see the machine boot properly from my new drive, and even more excited to see the vastly improved speed of everything. This four year old Mac feels like a new machine now.

Since there was no good articles online on how to remove an old Apple RAID card, I figured I'd share my experience here - in case anybody else out there is contemplating getting rid of their RAID card.

Obviously, the entire process would have been much smoother, had Apple actually supported USB 3.0 in that machine rather than FireWire 800 as the only high-speed external port. Even more important, a product like the Apple RAID card should never have been sold in the first place. It was poorly designed, suffered from battery issues, and slowed down the machine at random times outside of the user's control.

1 comment:

Richard Teusch said...

Be cognizant of read errors as they are a serious issue with RAID. If the disk fails, the RAID will have to read all the disks that remain in the RAID group to rebuild. A read error during the rebuild will cause data loss since we need to read all other disks during rebuild.

SATA disks experience read failure once every 12,5 TB of read operations, that means if the data on the surviving disks totals 12.5TB, the rebuild is certain to fail.