Perry The Cynic
Dabbler
- Joined
- Aug 15, 2023
- Messages
- 34
I've spent some quality time reading the docs and the forums. I'm now faced with needing to deploy some "converged" things that none of the repos seem to have (simple things like a container registry and bind9 support, and not-so-simple stuff like PostGreSQL...), and I'm a wee bit confused as to what deployment modes are supported, and how well. Please note that I'm not asking whether something will work. I'm asking whether IXSystems supports a particular deployment mode such that mistakes within it won't endanger the stability of the NAS and its core functionality, and regular upgrades won't turn my stuff into a shambles every time. (Obviously, if I provide the container, I'm responsible for its content.)
So here's how I understand the current state-of-play, i rough order of stability:
)?
At this point, my inclination is to use Official Apps where I can, use Docker containers where reasonably possible, and throw everything else into a single Debian VM that I maintain myself. Does that seem reasonable?
Cheers
-- perry
So here's how I understand the current state-of-play, i rough order of stability:
- VMs ("virtualization"): supported, and horrible disasters inside a VM shouldn't affect the NAS itself (except, obviously, where I'm connecting data or networking to it). Drawback: high basic overhead, I need to maintain an OS inside, limited access to NAS resources. Should survive NAS upgrades.
- Official App: supported. Should survive NAS upgrades. It's k3s but I'm supposed to consider that an implementation detail. Drawback: limited functionality, focus on stability over power.
- Docker containers: supported through the "Launch Docker Image" functionality. Shouldn't affect the surrounding NAS except where explicitly connected. Should survive NAS upgrades. Drawback: it's raw Docker.
- Community App: kinda supported, in a "well, it's a bunch of volunteers" kind a way. Might affect stability. Might break horribly during system upgrades, but IXSystems tries to keep things working. Drawback: It looks young and unstable.
- Build-My-Own-App (Helm chart conversion): Might work, but might affect stability. It's up to me to fix it if an upgrade breaks it. Drawback: Got to maintain my own app stuff for a client fleet of one; no clear stability guidelines (that I could find).
- Raw Helm chart: not supported. Might work fine, but system upgrades might just break it.
- Running stuff on the Linux that's inside SCALE: totally unsupported, potentially fatal to stability. Don't even think about it.
At this point, my inclination is to use Official Apps where I can, use Docker containers where reasonably possible, and throw everything else into a single Debian VM that I maintain myself. Does that seem reasonable?
Cheers
-- perry
Last edited by a moderator: