Hi,
just a slight correction for future references:
> They have to stay there unless you're willing to remove the jobs they
> ran, and that requires editing the core jobs database table by hand!
In Opencast's database, there is a connect between host registration,
service registration and jobs built with `on delete cascade` to easily
remove a host with all its services and jobs.
This means that dropping an entry from the host registration table will
remove the host just fine *but* it will also remove the services and
the jobs. This means that you loose all data about workflows run on
these nodes.
These data is still present in Elasticsearch though, meaning that just
deleting a host from the database means that you have inconsistent data
in your Opencast. To fix that, rebuild *at least* the solr workflow
index, the admin interface index and the external API index
(my recommendation: just rebuild all indexes).
In general, it's a good idea to just mess with the database if you
really know what you are doing. In this case, you could also just
disable the host using the REST interface which would have removed it
from the admin interface while keeping the data.
Best regards,
Lars