Wikimedia recipes should run only on Wikimedia VPS worker

[kelson42]

The reason is that these workers/servers benefit from a proviliged access (no request quota) to the Wikimedia API.

Today the assignement of taks to workers is based on worker based criterias:

• list of accepted scrapers
• memory/cpu limitations

Now we need a solution where part of the MWoffliner recipes have to be executed on part of the worker accepting MWoffliner recipes.

[kelson42]

One way to fix that would be to allow to tag/group workers AND create a new configuration "only on workers with tags" at recipe level.

[benoit74]

What is important to mention is that today the list of accepted scrapers and the resources allocated is defined in the worker itself. This means we have no control over this configuration in many cases since more and more workers are not under our direct control. Plus anyway it does not allow to "blacklist" one some recipes of a given scraper.

I like the idea of using tags at worker + recipe level.