Reload Nodes on Signal

[dennis-benzinger-hybris]

It would be nice if a reload of the nodes could be triggered by a signal. For example SIGHUP.

An example use case is a Rundeck Job which creates new nodes and then runs another job on these new nodes. After creating the nodes the first job would kill -SIGHUP chef-rundeck so that the second job sees the new nodes.

[leftathome]

Frankly I'd rather implement this as something over-the-wire, but if you're in a position to SIGHUP chef-rundeck you should also be able to dump its cached XML response file in its tmp directory. Have you tried that already? Did it give you the results you're looking for?

If not, then what we are looking at is a per-project and/or global cache dump endpoint. Or I guess we could support the PURGE command, Varnish style...

Do we know if Chef search results, partial or otherwise, honor ETags or any other cache-friendly headers?