Just woke up and made my coffee, found that my migration is complete!
I built a nas when I was fresh out of college with 3 drives in raid5 but because btrfs was relatively new at the time, I decided to go with ext4 for the file system. Essentially it defines how the 1’s and 0’s are arranged on the disk and how reads/writes etc work. Btrfs boasted some neat new features, but I just wanted reliable storage so I went with the established tech. I also left the 4th bay open because drives were expensive as frig at the time.
Now that time has passed and btrfs is more widespread, I found myself missing some of the features like snapshots and copy-on-write, so I decided to both add a hd and convert at the same time.
Only thing is that there isn’t really a way to convert an ext4 drive to btrfs - you have to copy everything off and back on. Some of the files I’ve got I’ve had for more than a decade, so I was understandably terrified of the thought.
I ended up making my new drive in the final bay into a standalone volume which I manually copied everything to, then wrote a bash script to recursively check the hash of every file on both sides to make sure they’re the same. This took 4 days of straight copying and reading/verification.
And cuz I’m a paranoid fek, I repeated the process with an old external drive I bought so I’d have two copies, one on the raid volume and on the external. Less chance of something going wrong on both simultaneously.
Even with two verified copies I had to take a deep breath and think real hard about anything I maybe forgot before I deleted that >10yo volume full of photos, legal docs, etc. Terrifying.
But this morning the restore has finished! Now I just verify the restored files and I can nuke the temporary standalone and add it into the raid cluster, and I’ll have a shiny new ginormous storage volume with all the lovely btrfs features I’ve been reading about.
The nice thing about this project is that I realized that this single drive failure that my raid cluster isn’t really the same as a full separate backup on a separate media. So I’ve left the cheap external that spins down if not accessed and set up a job to spin it up once a night and back anything new or changed to it.
This morning I’m sitting here excited about my objectively dull accomplishment. I think if I explained this to anyone I know irl their eyes would glaze over in the first paragraph, but I’m living my best boring life and it’s fantastic.
Btrfs is great, with copy on write by default and (offline) dedup and subvolumes/ snapshots. No need for LVM and reserving a fixed amount of disk space for a VM.
It adds a lot of complexity and I'm not sure what to do when it breaks (hopefully aunt google will help me then).