Ingest Node: Drop processor is ignored with on_failure

[spinscale]

Elasticsearch version (bin/elasticsearch --version): 6.5.1

Description of the problem including expected versus actual behavior: If a drop processor is used within on_failure of another processor, it seems to be ignored.

Steps to reproduce:

PUT _ingest/pipeline/pipeline1
{
  "processors": [
    {
      "fail": {
        "message": "failed",
        "on_failure": [
          {
            "drop": {}
          }
        ]
      }
    }
  ]
}

PUT foo/bar/1?pipeline=pipeline1
{
  "foo": "bar"
}

# this document exists, but should not
GET foo/bar/1

This example looks somewhat artifical, but it might make more sense with another if in the drop and maybe a grok processor instead of fail - I just picked those to shorten the example.

original issue in the discuss forum

[elasticmachine]

Pinging @elastic/es-core-features

[ajazam]

I am the person who originally created the forum issue. I had to sanitize the code to remove company specific information.