I sure do hate having to post here on these terms (again, kind of) but this is only the beginning, I’ll keep this thread concise.
After lots of help from folks here, many tedious hours of work, I almost had my NAS/Jellyfin/etc all back up after a RAM failure. Fast forward through a ton, don’t wanna talk about it.. a LOT of stupid decisions later (trusting Gemini to give me instructions) it somehow led me to completely brick my server again. I’ll save more detail on that for another post because I still want to set up a Bazzite VM, and yes I’m an idiot. But for now..
I am now reinstalling SCALE mostly from scratch AGAIN. Here is an issue I ran into the last time I did this, which I feel there must be a better way to tackle than how I did:
When you have to build your datasets over again, and your ZFS pool has folders with the same names/locations as the dataset you’re trying to create, it gives you errors. For example, I restored as many files as I was able to backup to Google Drive previously, then back onto my ZFS pool. Say there is a dataset I create called “media,” and inside it, all of my restored Google Drive content, one folder is called “tv” and one is called “movies” etc. When I tried to make a dataset called.. well let’s start with “media/tv” because there is already a folder by that name, I get an error saying there’s already a folder by that name in that location. Gemini (dumb) told me to move the folders to another place with mkdir / mv, create the dataset, then move them back in, using the TrueNAS shell. Is that really the best way to do this? I read another thread that made it SOUND like there is a better way (renaming, I think?) but can’t find it now.
I’m scared of using the TrueNAS shell at this point after the nightmare I’ve been through recently. But will trust folks here.
A lot of preamble for what seems like a pretty simple question, sorry I tried. Not good at brevity.
Why do you need an independent dataset underneath media? For my Plex installation, I just use folders underneath my media dataset.
I wrote an article on the old forum on how to structure datasets in a pool:
Basically, only create datasets when you need to:
You intend to have separate snapshot or replication intervals or retention periods.
You’ll need to consume them in different apps or containers
You’ll need to assign different permissions or ACLs for security reasons
etc. You get the idea. In your case, everything in your media dataset is for Jellyfin, and will presumably have the same snapshot intervals & retention. I don’t see a need to create daughter datasets underneath your media dataset. Jellyfin won’t know about or care the subfolders are or aren’t datasets, so long as the paths don’t change.
Ehhh I am a complete idiot when it comes to this, despite being very “good” as a youth with PC’s and nearly completing my CCNA.. TrueNAS and anything Linux related has scrambled my brain. So now with a fresh head I’ll clarify:
I did pointlessly create separate datasets for the daughter folders in the past, yes. And thanks for making it clear that that was unnecessary. However, I also have “configs” and “stacks” folders and within them various datasets for: Radarr, Sonarr, Qbittorrent, Prowlarr, Recyclarr, Jellyfin, Jellyseerr, Bazarr.. maybe some others I’m forgetting. I also have folders for them already in place in the ZFS pool, as the pool remained in tact between OS breakdowns/re-installs.
And now as I write this, my ZFS pool has actually become corrupted AGAIN and needs to be completely wiped anyway.. FUN TIMES. BUT for future reference, in this case where the Host OS needs to be reinstalled but the ZFS pool remains intact, is the best thing to do really to cut/paste these folders to a different location, create the datasets, then cut/paste them back? Or am I understanding this wrong.. is it that the datasets should have been saved and configured in the backup configuration file of TrueNAS Scale and not require any further intervention?
Excuse my confusion and ignorance, I’ve had every problem I can imagine with my server over the last 6 months, rapid fire back to back, and it’s worn me down. Working on another post about the Bazzite VM now :’(
I have nothing useful to add but my sympathies. Once you get it all working again, you’ll be halfway to a pro though! (the other half being the ‘getting paid’ part of a professional)
Your dataset/folder/data structure have nothing to do with what is saved in the config file. The data pool is living independently from your config and entirely self-sufficient on your data discs.
