Last week, a second hard drive in my Promise Pegasus R4 RAID array died. The first had given up the ghost late last year, and had been duly replaced. The immediate question was what to do once I had turned Time Machine off, and shut the array down.

When I hand-build RAID arrays, or any other multi-drive systems, I like to use identical hard drives from different batches. The reason is that hard drive life varies considerably between batches, but those within a given batch tend to fail at around the same time. Inevitably commercial manufacturers like Promise tend to assemble RAID systems using drives from the same batch.

Now that two of the four drives had failed within a couple of months, I knew that the chances of the third and fourth drives failing soon were high. So simply replacing the failed drive would be a temporary expedient.

At the same time, using the RAID array just for Time Machine backups, it was running out of room. I think that the earliest backup dates from 2012, and there was only around 450 GB remaining out of the 3 TB storage space provided by the four 1 TB drives (this being RAID level 5). What I really wanted to do was to swap its 1 TB drives for 2 TB units, which would give me a net total of 6 TB to be working with.

In an ideal world, I would have located a 3 TB Thunderbolt drive, copied the whole backup set onto it, upgraded the hard drives in the RAID, then copied the backups back onto them. Although time-consuming, it would have worked.

With only three of the four drives working, I needed to rebuild the RAID array back onto the fourth drive before I did anything else. It seemed daft to do that and then mess around upgrading its capacity, so I am trying to do both together.

I ordered four 2 TB Ultrastar ‘Enterprise spec’ hard drives compatible with the Promise hardware, which arrived today. I first removed the failed drive from the enclosure, replaced it with one of the new drives, re-inserted the drive module, and started the Pegasus unit up. It was not set to rebuild arrays automatically, so using the Promise Utility I set it to manually rebuild the new drive.

Once that is complete, I will shut the Pegasus down again, pop the next 1 TB drive, swap in one of the new 2 TB drives, and rebuild the array once again. I will repeat that until all the four drives are new 2 TB units.

At that stage, the chances are that the logical drive will remain set as the original 3 TB RAID 5 which was provided by the original 4 TB of physical drive space. I will then try to increase that to 6 TB, using Promise Utility’s Migrate feature (which can also be used to change RAID level). If that works, I should have the free disk space which I need, and drives which should last me another 5 years or so.

If migration does not work, then I will be forced to do what is perhaps the longer-term option anyway: to wipe the array, create a fresh 6 TB logical drive, and start Time Machine from scratch.

I will let you know how I get on.