I'm planning to replace my 4TB drives in a RAID 1+0 array with 4 new 22TB drives. To avoid risking data during the transition, I want to add one of the new drives as a hot spare, fail over an old drive to it, and then gradually replace the old drives without dropping below mirror redundancy. My current RAID setup consists of two sets of mirrored drives, and I want to ensure that while I replace the drives one by one, I don't expose myself to the risk of a drive failure that could wipe out my data. What steps can I take to achieve this safely?
5 Answers
You can definitely expand each of your RAID 1 arrays to include a third drive temporarily. This way, you'll have a backup mirror while you work on the replacements. Use these commands to add a new drive and then fail the old one:
`# mdadm /dev/md*XX* --add /dev/sd*Y*`
`# mdadm /dev/md*XX* --fail /dev/sd*Z*`
`# mdadm /dev/md*XX* --remove /dev/sd*Z*`
If you're cautious, you could space out these commands over a couple of hours. This method provides a safety net without compromising your array. Plus, check `/proc/mdstat` for a clearer status of your RAID setup if you need it!
Exactly! Just make sure the new drive syncs properly before you fail the old one. That way, you keep everything intact throughout the process. It might seem complicated, but taking it step by step will serve you well.
What's the capacity of your new HDDs?
Each of the new ones is 22TB! That's a huge step up from the 4TB originals. Taking it slow on the integration is key for me.
At this point, you might want to explore using BTRFS or ZFS instead of sticking with RAID. They can offer simpler management and might even better accommodate your future needs.
I've been using ext2 for ages! I'm curious about BTRFS and ZFS since I've heard people rave about them. I just want to know if they fit my requirements.
I recommend testing this process with smaller loop devices first to make sure your plan works as intended. It can help minimize risks when you're handling the real drives. It seems with RAID 1+0 you might not be able to simply add and remove drives as freely as you'd like, so it's good to verify your approach first.
That’s smart! I can definitely try a lower risk version first. I want everything to run smoothly with the new setups.
Before starting your upgrades, if you haven’t done so already, consider creating an offsite backup. If the new drives aren't yet in use for backups, you can utilize them for redundancy during the transition. Safety first!
Yeah, I'm guilty of slacking on real backups. I'm just waiting on a larger drive to really set that up. But I still want to replace the drives one at a time even with the backup in place.
So, doing it this way ensures the new drive stays live on the mirrored sets? I just want to be clear that the new drives are utilized while the old drives are taken offline. I certainly want to avoid downtime since I deal with frequent power fluctuations.