Kimmo Lehto

Results 287 comments of Kimmo Lehto

It's something else. The setters don't care where the value came from. Are you perhaps using a nonexisting grid?

It's built by using this monster: https://github.com/kontena/kontena/blob/fceb2942cef53b58261aea80600c3887631f1723/cli/lib/kontena/cli/services/services_helper.rb#L42-L215

I think it should be [/run/k0s.pid](https://refspecs.linuxfoundation.org/FHS_3.0/fhs/ch03s15.html) (altough `/var/run` likely is a symlink to `/run` on many systems)

[OpenRC service script docs](https://github.com/OpenRC/openrc/blob/master/service-script-guide.md): ![image](https://user-images.githubusercontent.com/224971/174601642-c9a9f502-2ccc-4915-ac1a-4574d466bc54.png) ![image](https://user-images.githubusercontent.com/224971/174601711-6947cf84-1cbc-4192-aff1-ce858d72c49c.png) -- Looks to me like OpenRC really wants a pidfile It can create one on its own though: -- ![image](https://user-images.githubusercontent.com/224971/174601875-878d5700-2755-4a79-8d04-ece8d31af06f.png)

This would probably help when setting up to run k0s as non-root. (maybe that should have its own documentation too? are there caveats? can't run privileged containers?)

k0sctl relies on using the `--help` of subcommands to figure out if the k0s version has that subcommand or not.

Do we have adoption stats?

Thinking if it would be feasible to change all file/path references in k0s code to some kind of `NamedPath` objects that could be used to generate docs (or code would...

> windows support is experimental, may change later, and even use windows, there must have a controller+worker run in linux Yes, but we do have several `FooFooDefaultWindows` paths in the...