Prometheus Integration not scraping summary metrics

I opened a Github issue (Add support for summary metrics, when collecting from Prometheus exporter · Issue #37376 · elastic/beats · GitHub), which didn't any responses, so I will repost here. I opened it as enhancement issue, but I assume it is a bug, so please ignore the following template for enhancements:

Describe the enhancement:
We are currently trying to use Elastic Agent and Metricbeat in connection with the Prometheus/Prometheus Input integration to scrape metrics from Sonatype Nexus3 Repository Manager (which provides a prometheus formatted metric endpoint Support Features).

Some of the metrics are formatted like the following:

HELP org_sonatype_nexus_blobstore_file_FileBlobStore_get_timer Generated from Dropwizard metric import (, type=com.codahale.metrics.Timer) TYPE org_sonatype_nexus_blobstore_file_FileBlobStore_get_timer summary org_sonatype_nexus_blobstore_file_FileBlobStore_get_timer{quantile="0.5",} 5.9E-6 org_sonatype_nexus_blobstore_file_FileBlobStore_get_timer{quantile="0.75",} 0.0060767 org_sonatype_nexus_blobstore_file_FileBlobStore_get_timer{quantile="0.95",} 0.009478 org_sonatype_nexus_blobstore_file_FileBlobStore_get_timer{quantile="0.98",} 0.009478 org_sonatype_nexus_blobstore_file_FileBlobStore_get_timer{quantile="0.99",} 0.009478 org_sonatype_nexus_blobstore_file_FileBlobStore_get_timer{quantile="0.999",} 0.009478 org_sonatype_nexus_blobstore_file_FileBlobStore_get_timer_count 481958.0"

When using the prometheus integration to scrape the whole endpoint, the metrics with types like summary are not indexed in Elasticsearch. If this is a bug, please convert the issue to a bug issue.


18:21:20.133 elastic_agent.metricbeat [elastic_agent.metricbeat][info] Non-zero metrics in the last 30s 18:21:34.324 elastic_agent.metricbeat [elastic_agent.metricbeat][info] 'ca_trusted_fingerprint' set, looking for matching fingerprints 18:21:34.324 elastic_agent.metricbeat [elastic_agent.metricbeat][info] CA certificate matching 'ca_trusted_fingerprint' found, adding it to 'certificate_authorities' 18:21:34.327 elastic_agent.metricbeat [elastic_agent.metricbeat][debug] Completed dialing successfully 18:21:34.327 elastic_agent.metricbeat [elastic_agent.metricbeat][debug] Error reading from connection: read tcp AGENT_IP:PORT->ELASTICSEARCH_IP:9200: use of closed network connection 18:21:34.336 elastic_agent.metricbeat [elastic_agent.metricbeat][debug] PublishEvents: 1 events have been published to elasticsearch in 21.2452ms. 18:21:34.336 elastic_agent.metricbeat [elastic_agent.metricbeat][debug] ackloop: return ack to broker loop:1 18:21:34.336 elastic_agent.metricbeat [elastic_agent.metricbeat][debug] ackloop:  done send ack

Maybe the above shown error has something to do with it (maybe metricbeat is running in a timeout?).

Describe a specific use case for the enhancement or feature:
If you index metrics into elasticsearch, you do not want to get only half of them. Summary types should also be supported.

My version is Elastic Agent 8.11.3 with the Prometheus Input Integration version 0.2.0

still relevant