2026-09-29, 11:48 AM
(This post was last modified: 2026-09-29, 11:49 AM by TheDMV. Edited 1 time in total.)
https://github.com/TheDMV2/Jellyfin_Dock..._Migration
In short: I moved from Docker on a Synology NAS to a 16GB M1 MacBook Air. I got AI (mostly Claude) to help me do it, and every step I'd find something else missed that would need to be fixed. But eventually everything was migrated properly over, and it's working great, with nothing lost.
I figured, given how cheap M1 hardware is, other people would probably like some help too. (M1 MBAs with broken screens can be found for about $200USD, fairly easily, for example). Plus, it handles basic AI stuff easily enough, and Orbstack + Portainer isn't ideal, but it's still a lot faster than on my NAS.
So, after it was all working, I told the AI to collate everything we did and turn it into a single how-to document. If you don't like pretty scripts, just follow that.
Because it's fairly straightforward, and just a lot of scripting, I got AI to turn all of that into a single script. Just edit the first section with all your paths/libraries, and it will do the rest, safely.
* it backs up
* it does a dry-run (no edits) first
Testing (of the script): minimal. Worst case, you can just delete and copy everything back and go again, since you still have your docker setup which is never touched.
If anything is weird/broken, let me know.
To be clear: this is ONLY for Jellyfin 12.x - due to database structure and naming changes, and that I no longer care about JF 10.11.11 and below, the script is hardcoded to abort on non-12 installations.
https://github.com/TheDMV2/Jellyfin_Dock..._Migration
*[AI included in results/code, not in this text]*
-----
By the by: Macbook Airs tend to be cheaper than MacMinis, because Minis are being grabbed by people for AI coding and such.
Personally, I prefer using the MBA, since it basically has a built-in UPS. I modified mine with the thermal hack (where you open it up and put some thermal pads to help turn the bottom of the laptop into a headsink, allowing it to run hotter, longer. And no, this doesn't heat up the batteries; they're too far away and not thermally touching). That was mostly for some AV1 transcoding stuff I was doing; as JF server, it doesn't need this at all. I also grabbed a 2.5Gbps USB-C ethernet dongle for mine (and disabled wifi), just because I could. Obviously, this is something that the Mini doesn't need.
Weirdly, MBA and Minis have different power settings in the OS. In short, Amphetamine/Caffeine apps work great to keep the laptop on while closed.
If you're going to run Ollama on your Macbook like I do to handle Mealie and Paperless requests, it's okay, because local AI pushes the GPU, while Jellyfin pushes the CPU, so they don't necessarily directly conflict with each other when running hard.
In short: I moved from Docker on a Synology NAS to a 16GB M1 MacBook Air. I got AI (mostly Claude) to help me do it, and every step I'd find something else missed that would need to be fixed. But eventually everything was migrated properly over, and it's working great, with nothing lost.
I figured, given how cheap M1 hardware is, other people would probably like some help too. (M1 MBAs with broken screens can be found for about $200USD, fairly easily, for example). Plus, it handles basic AI stuff easily enough, and Orbstack + Portainer isn't ideal, but it's still a lot faster than on my NAS.
So, after it was all working, I told the AI to collate everything we did and turn it into a single how-to document. If you don't like pretty scripts, just follow that.
Because it's fairly straightforward, and just a lot of scripting, I got AI to turn all of that into a single script. Just edit the first section with all your paths/libraries, and it will do the rest, safely.
* it backs up
* it does a dry-run (no edits) first
Testing (of the script): minimal. Worst case, you can just delete and copy everything back and go again, since you still have your docker setup which is never touched.
If anything is weird/broken, let me know.
To be clear: this is ONLY for Jellyfin 12.x - due to database structure and naming changes, and that I no longer care about JF 10.11.11 and below, the script is hardcoded to abort on non-12 installations.
https://github.com/TheDMV2/Jellyfin_Dock..._Migration
*[AI included in results/code, not in this text]*
-----
By the by: Macbook Airs tend to be cheaper than MacMinis, because Minis are being grabbed by people for AI coding and such.
Personally, I prefer using the MBA, since it basically has a built-in UPS. I modified mine with the thermal hack (where you open it up and put some thermal pads to help turn the bottom of the laptop into a headsink, allowing it to run hotter, longer. And no, this doesn't heat up the batteries; they're too far away and not thermally touching). That was mostly for some AV1 transcoding stuff I was doing; as JF server, it doesn't need this at all. I also grabbed a 2.5Gbps USB-C ethernet dongle for mine (and disabled wifi), just because I could. Obviously, this is something that the Mini doesn't need.
Weirdly, MBA and Minis have different power settings in the OS. In short, Amphetamine/Caffeine apps work great to keep the laptop on while closed.
If you're going to run Ollama on your Macbook like I do to handle Mealie and Paperless requests, it's okay, because local AI pushes the GPU, while Jellyfin pushes the CPU, so they don't necessarily directly conflict with each other when running hard.