Make the roles.json and nodes.json files public

[gravitystorm]

The readme says "Currently the _data/nodes.json and _data/roles.json are not publicly available."

This makes it impossible for anyone to help develop this codebase. When I was working with those files, I didn't see anything that I thought was sensitive information, but I might easily have missed something. Any thoughts @tomhughes?

[tomhughes]

Well roles shouldn't be an issue, because it's all in the public repo anyway.

Nodes is pretty much impossible to verify I fear, and to be sure that something random won't change on a chef upgrade (or more precisely an ohai upgrade) that could wind up leaking something sensitive.

[abitrolly]

Can there be an example of these files at least?

[tomhughes]

Not easily - the node data is a huge amount of automatically generated data for each machine combined with specific customisation from the cookbooks. You can't really write a dummy node.

[abitrolly]

https://github.com/gravitystorm/osmf-server-info/blame/master/index.md uses only a subset of data. Is it possible to filter this node.json to export only these?

• node.automatic.roles
• node.name
• node.automatic.ohai_time
• node.default.hosted_by
• node.default.location
• node.override.country

Looks like there is no other info being used.

[tomhughes]

That's just not true - that might be all the index uses but other pages use a lot more.

What exactly are you trying to do? I'm not really sure why anybody that is an OSM system administrator would want to make a change here? Pretty much anything you do here needs fairly deep knowledge of our infrastructure and often a lot of experimentation to find (or make available) the necessary details.

[abitrolly]

@tomhughes need the same feature for the local community - osm-by/OpenStreetMap.by#17

[tomhughes]

So are you using chef to manage your servers? or at least gathering data from them with ohai? Because if you're not then I doubt this repository is going to help...

[abitrolly]

@tomhughes I don't think that Chef is a suitable model for the servers owned by different people who may want to run their own payload there as well. I think about them gathering data and sending them to some endpoint, with ohai or Ansible facts, or some kind of JSON version of neofetch.