@zaidka What about this one? ;-)

Won't this cause cache to clear whenever a new worker process starts?

I'm not sure. I verified the execution of the clear_permissions_cache method with log output. And the log output appeared only once. I will have a look again tomorrow.

I had a look at this once more. The clear_permissions_cache method is called per worker. This means that it is called more than once at startup if you run puma in clustered mode.

There's another possible place to clear the cache on startup, in config/puma.rb:

before_fork do
  # action
end

But I don't have access to the Rails or the GenieacsGUI object there AFAIK.

@zaidka Do you have an idea how to fix this?

Maybe clear the cache manually from console or using a rake command.

Well, then I'd rather have the disadvantage of getting the cache cleared too often than having to run a command manually 😁

Why not cache in memory instead?

You mean using MemoryStore as the cache store? (https://guides.rubyonrails.org/caching_with_rails.html#activesupport-cache-memorystore) Well, that's an option too, but then there's a cache per worker as I have understood it.

Should we go this way?