rsvg 14 Posted February 24, 2021 Share Posted February 24, 2021 Hi, I'm interested in being able to run multiple instances of Emby in my k8s cluster. This doesn't seem to be possible because the sqlite database becomes locked when one instance is running. Without offense, this seems like a generally outdated data storage method, and is inherently unfriendly to modern scalable architectures (even when those are scaling on small home pcs). Is there any plan to allow a better database such as SQL or Postgres in the future? I'm aware that the codebase is not open source any more, or not completely? Is a fork of this utilizing a remote database possible? As kubernetes grows in popularity for simple HA, I think this is definitely going to be of interest for many people and would love to help support it. Thanks. 1 Link to comment Share on other sites More sharing options...
PenkethBoy 2063 Posted February 24, 2021 Share Posted February 24, 2021 there is a feature request for this - lend your support to that - but its old and i guess unlikely to happen - basically it would require a major re write and with fast cpu's and ssd's these days not needed - so I cant see the need for multiple instances unless they have different data - and you can already run multiple instances of emby for that purpose. Link to comment Share on other sites More sharing options...
ebr 14912 Posted February 24, 2021 Share Posted February 24, 2021 Link to comment Share on other sites More sharing options...
kikinjo 162 Posted February 24, 2021 Share Posted February 24, 2021 For any really HA setup i agree with rvsg, database is limiting factor, but not only db, currently now you can only have 1 emby server with ssl setup (or with cloudflare). Yes some reverse proxy is possible of course but that's added complexity and more layers for problems. Emby is way behind if we talk about real HA setup..or scenario where someone would like to run it in Active : Active / passive setup for 2 emby servers. Link to comment Share on other sites More sharing options...
rsvg 14 Posted February 24, 2021 Author Share Posted February 24, 2021 Quote so I cant see the need for multiple instances unless they have different data Perhaps I can help you see the need. You have a small, cheap home machine and begin building your collection. Then your family grows and your machine can no longer handle the transcoding if multiple users are on at the same time. It could be more scalable, and cheaper to add another small cheap machine than upgrading your whole computer to a newer one. This is a very common separation of concerns. The database shouldn't be bound to the instance. Cattle not pets. Link to comment Share on other sites More sharing options...
ebr 14912 Posted February 24, 2021 Share Posted February 24, 2021 4 minutes ago, rsvg said: Perhaps I can help you see the need. You have a small, cheap home machine and begin building your collection. Then your family grows and your machine can no longer handle the transcoding if multiple users are on at the same time. It could be more scalable, and cheaper to add another small cheap machine than upgrading your whole computer to a newer one. This is a very common separation of concerns. The database shouldn't be bound to the instance. Cattle not pets. Please carry on this discussion in the existing feature requests but much of this has been discussed there already. It boils down to something that is very interesting to us but not really all that necessary for the vast majority of our users in our current target market - especially with our ability to use graphics cards for transcoding muscle. If and when we did this it would likely be a new class of product targeted at something like an enterprise. 1 Link to comment Share on other sites More sharing options...
Recommended Posts