Do you still have to use the ancient Docker version 2.5.0.1 on the Mac, or is there a newer Docker version that works well with Storj?
I strongly do NOT advise to host a node on MacOS. You will most probably have issues with docker and file system…
Hello @NeoA,
Welcome to the forum!
This is the last stable version of docker for Windows and macOS, which having less issues.
However, you may try to run the latest and report back - do you have issues or not.
We pinned the version based on SNO’s feedback.
Please also note - Mac M1 likely will not work at all or will be incredible slow (M1 doesn’t have hardware virtualization, required to run docker under macOS, so it will use software emulation). For that version of Mac you must use the special docker version: Install on Mac | Docker Documentation
Why not at this point just run it natively, via launchd? I did not check if there is storagenode binary for macOS/Darwin but go compiler definitely exists there and these binaries can be made with minimum efforts, both for apple silicon and amd64.
Running monolithic go binaries in docker, (that has to run in another abstraction layer -vm, by necessity), on a *BSD derived modern OS is a very questionable approach to say the least.
Docker only brings issues and provides zero value at this point, especially for storagenode.
A this point it simply reduces to an environment to download, start, and stop software. And yet each os has their own tools for that — launchd, task scheduler, systemd, rc…
Next step would be write a simple wrapper that would ask user questions and do all configuration described in SNO onboarding for them. Will generate service descriptions on each OS and massively simplify deployment. And then bye bye docker and associated issues, overhead, and headaches.
Ha! Just checked. There are both arm and amd64 osx binaries. So, why are we still discussing docker? And why did I have to google “GitHub releases storj” to find them, as official documentation only links to docker, as if it is the only way?
It’s a supported way. You can use a binary too, but there is no step-by-step guide.
However, it’s relatively simple, if you know how to run a binary on start and how to type commands in the terminal. Docker just convenient delivery method and it’s a cross-platform.
Right, but running docker still requires typing commands in the terminal 
Maybe fewer commands, but still, there is no single “run storj for me” button
Indeed, but when we end up cherry-picking docker versions to walk on eggshells around various docker-related bugs – cost of that convenience is no longer justifiable.
Even on Synology, where DSM provides a “nice” UI for docker, people often end up running software natively just because of these self-inflicted avoidable issues imposed by the containerization. And that is linux, where docker runs in a user space. On Win and macOS this requires a VM. So it becomes macOS → VM → Linux → container → linux → storj. Instead of macOS → storj. And extra typing can be mitigating by STORJ providing official daemon configurations, e.g. launchd plists for macOS. I’m sure users will be happy not needing to download and run 1.5GB monster just to run a very lightweight and efficient storage node software
We have no capacity to support GUI versions for all platforms. But it’s an Open Source and the Community contribution is very welcome!
2 Likes
I’m not strong enough now to do other undocumented installs. my old Docker version just crashes more often lately. I have this running on Macmini late 2012, on Macos Mojave. Catalina was too buggy. I thought maybe someone already has experience with newer Docker versions and could report. The problem is, if you update to the latest Docker version, for example, the containers are also adjusted and it’s not so easy to come back.
It’s actually easy, all containers are ephemeral. So uninstall the new version and install old version back, then run your container with all your parameters.
I’m running Docker 4.8.0 (78933) now. this is the last version supported on Macos Mojave. so far it’s going well.
1 Like