@cachedout I've never built an LS plugin and tested it in conjunction with another LS core PR. Would you mind posting some instructions on how I can essentially run this PR along with elastic/logstash#10561? Thanks!

@ycombinator I did my development by editing the installed gem directly, the instructions here should allow for testing in your environment after checking out this PR into a local branch: https://github.com/logstash-plugins/logstash-output-elasticsearch/blob/master/README.md#2-running-your-unpublished-plugin-in-logstash

Relevant PR on Logstash core is elastic/logstash#10636

With it, the implementation of LogStash::Outputs::ElasticSearch::Common#discover_cluster_uuid becomes something like:

    def discover_cluster_uuid
      if defined?(plugin_metadata)
        cluster_info = @client.get('/')
        plugin_metadata.set(:cluster_uuid, cluster_info['cluster_uuid'])
      end
    end

And extracting this metadata in your other PR becomes something like:

  def resolve_cluster_uuids
    outputs.each_with_object(Set.new) do |output, cluster_uuids|
      if LogStash::PluginMetadata.exists?(output.id)
        cluster_uuids << LogStash::PluginMetadata.for_plugin(output.id).get(:cluster_uuid)
      end
    end.to_a.compact
  end

jenkins, re-test this

The failing tests also appear to be failing on the master branch for me.

I had to also fix up some other tests which were failing because they depended on a certain sequence of log events instead of being fully isolated.

@yaauie can you re-review this?

Changes pushed. This is still failing a local test but I believe this is a problem unrelated to this PR. It also fails in the master branch and is due to a problem that occurs when attempting to fetch the ES version with a path specified.

As such, I believe this is ready for merge.