I have a jellyfin server running debian 11 that was modest spec when I built in early 2013, and isn’t doing a great job of keeping up with new media encoding or multiple streams now.
I’m looking to upgrade the core components but the thought of reinstalling the OS, rebuilding the services, and reconfiguring everything has me putting it off. It’s running an AMD A6-5400k APU, the new target hardware will be Intel i7-7700k w/ a dedicated GPU. There’s also 4 storage HDDs and a small 128gb system SSD for the host attached.
Ive heard Linux ships with drivers for most everything it supports… so I’m wondering If I were to just swap the mobo/CPU/ram could I reasonably expect debian to “just work”, or are my hesitations right and I need to plan to reinstall the OS? Is there any way to back up the system and configuration to simply restore it after the new hardware is in? I’d really like any suggestions to make this as painless as possible! Thank you
I ran jellyfin in a VM. I literally just exported and imported it. All I had to do was change the UUIDs of the storage disks.
I recommend getting the new server as a whole box, install everything and configure as needed. When everything works as you require, switch feeds. Higher cost but minimal downtime, low risk.
Do you even need gpu? I have much cheaper g3930 and I can stream multiple 1080p at once. Not sure about 4k tho, but quicksync is amazing
I have an inherited GTX970 that I thought maybe could put to use, I do agree the i7 w quicksync would probably be okay though. We’re hoping to support 3 simultaneous 1080p clients
I recommend suck it and see. Last year I migrated from an intel 4770 to a AMD 5700X with a nvidia 980 in tow. Out of morbid curiosity I tried booting it (Fedora) before re-installing and it just worked, installed microcode and it was fine.
Bite the bullet, backup everything and start fresh. This time, setup LVM (w/wo encryption, your choice) and make sure to setup /var, /tmp, /home, etc. as separate volumes.
In another 10 years when you need to fully upgrade, you can preserve specific volumes and just swap out the system volumes with new installs. This also helps in case of issues which require system refreshes. Most importantly, as long as the drive isn’t busted, all you have to worry about backing up are specific configs and your home volume. The rest becomes simple.
I would recommend running a couple disaster recovery scenarios once setup.
If I were you I’d try increasing RAM and adding a dedicated GPU or two for transcoding work… they don’t need to be the latest and greatest to have h.264 built in.
This is an interesting thought, I already got the i7 + DDR4 ram for cheapcheap under assumption the CPU would be a bottleneck, but I probably could find some used DRR3 and bump that + the gtx970 🤔 sounds less risky than a mobo swap
why a 7700k? I think something like a 12100f has better performance and much lower power draw, meaning much less cost in the long run. It’s cheap as well.
Yes agreed, it is not ideal for power consumption. I found the i7+Mobo+16GB DDR4 at a flea market for $120 and just jumped on it. This was a couple months ago. I knew it needed an upgrade but didn’t want to spend $300+ and bundle was hard to pass up 😁 if I was buying new definitely would have gone with i3.
I do somewhat regret getting LGA 1511 as there is no upgrade path for this mobo.
I see, yeah $120 is a great price. I would buy that in a heartbeat as well
It is possible and for backup use timeshift
Hmm time shift is whole system backup right? I do have root system backups with rsync on a systemd timer, but if the hardware changes and doesn’t work with current system, the backups will probably be useless as well? I will read more about timeshift thank you
I would not recommend swapping out major components without doing a new install. Linux installation is quite simple these days, so back up your configs and put them in place in the new installation.
If you run things in containers, these kinds of issues are not as relevant. You can run jellyfin in docker the same on any host and pass the GPU through for xcoding.
Unfortunately I don’t think there’s enough info to answer your question specifically.
I don’t use containers because I haven’t wrapped my head around docker lol. But I have thought maybe that portability could be worth implementing with the new hardware… right now it’s just the host system with jellyfin, a few *arr apps and supporting services installed through apt with their recommended repositories
Thank you though I will consider this approach. If I moved to docker, setting it up may be worth it to avoid this headache next time
So for a cheat code, check out “ansible-nas” on github. It will do all the heavy lifting for you.
Waahh this looks incredible. You have provided much to think about, this seems perfect