Organization runners

[MPV]

Hi,
Interesting blog series, thanks @wayofthepie

Are you running this today, or are you using something like https://github.com/evryfs/github-actions-runner-operator/ or https://github.com/summerwind/actions-runner-controller/ nowadays?

Also, what's your thoughts on trying this for org-level runners?

[wayofthepie]

Hi @MPV, thanks!

Im not currently using self-hosted actions, we are able to get away with the public ones where I work. However this will be necessary soon enough and we will likely go with a kubernetes based approach. I wouldn't use what I have in this repo for anything other than exploring the API's 😄 it was just a way for me to better understand how the runners work.

I wish they exposed more of the internal workings, so we could build on them outside the actions-runner repo. For example if you look at a runner's config, the .runner file, it has a field serverUrl that looks something like https://pipelines.actions.githubusercontent.com/xcH99UdnlYnfqDyvmge4GBRq/. If you call this with the right credentials it will hang and time out after 60 seconds. But, this is where run request events get sent for a specific runner, so if you long poll it and test with a new build you will get an event. If this was documented better and was deemed a public API you could build true event based runners using it.

Technically you can build this now but because its not really documented, its just an implementation detail of the runners, it could break at any time. The org level runners make this an even more powerful solution. Thinking about all this again makes me want to build it out properly 😄