We have two new dumpsdata hosts to be deployed, and we're tight on space on the dumpsdata fallback host we have, since it gets rsyncs of XML/SQL and misc dumps. As more hosts do wikidata jobs after they are done with the regular jobs, we can expect pressue on disks and network to go up too. Additionally, the procedure for swapping a dumpsdata primary and secondary is very fragile. What can we do to fix this all?
An earlier version of this task was dropped when the issue it was trying to address (super slow rsyncs to public) was fixed: T254856