In the case that the plugin gets a log entry from a resource that no longer exists and does not exist in the cache, the Kubernetes API will return an error stating that the resource was not found. This slows down processing considerably.

This can happen if logstash is restarted before it has caught up and a node that it is processing logs for no longer exists. If the cache were persistent, this would usually be OK but we also should cache the fact that the resource was not found if it was not already in the cache so we don't slow down processing.

Also, if the node does not exist and we can't find info about it, we should make sure we still write the node name to the data as we do know that much.

Looks like the Logstash team doesn't have permission. I get this:

You do not have permission to push to this gem. Ask an owner to add you with: gem owner logstash-filter-kubernetes_metadata --add [email protected]

The current documentation exists embedded as comments in the plugin. This is no longer used for plugin docs. Instead, we recently moved to putting the documentation in docs/index.asciidoc

Example: https://github.com/logstash-plugins/logstash-output-elasticsearch/blob/master/docs/index.asciidoc

@ph has a tool to extract this automatically, but it can be done manually if desired.