Unable to fetch prometheus c++ exporter stats on metricbeat

[krishan-agrawal-guavus]

Prometheus C++ exporter was modified to export our application stats and we planned to export these stats to Metricbeat (and to ElasticSearch) on some remote server.
My issues is that I am able to get Prometheus stats from my Kafka cluster using JMX exporter configuring the correct port details.
However, when using the cpp exporter I have not been able to get these stats into Metricbeat.
Although, when I do request using browser or curl request on configured port, then I can see those stats.

Do I need to take care of any pre-requisites while exporting Prometheus stats using this c++ exporter Or any specific conditions to be met so that metric beat can read and export these stats?

Sample exporter stats:
collector_stats{status="YYY",type="XXX"} 0.000000

[exekias]

Hi @krishan-agrawal-guavus, thank you for your feedback! We will need some more info in order to debug this issue. Could you share these?:

• Metricbeat settings you are using
• Metricbeat logs when running it
• The full dump you get when manually fetching the metrics from the C++ exporter

Best regards

[krishan-agrawal-guavus]

Metric beat settings:

1. Enable prometheus and system stats
2. The output of prometheus.yml:

- module: prometheus
  period: 10s
  hosts: ["localhost:8090"]
  metrics_path: /metrics
  namespace: example1

The version is 6.4

The logs of metricbeat are attached.
metricbeat.gz

Here are the dump of metrics from the C++ exporter

# HELP exposer_bytes_transferred bytesTransferred to metrics services
# TYPE exposer_bytes_transferred counter
exposer_bytes_transferred 17701173.000000
# HELP exposer_total_scrapes Number of times metrics were scraped
# TYPE exposer_total_scrapes counter
exposer_total_scrapes 16166.000000
# HELP exposer_request_latencies Latencies of serving scrape requests, in microseconds
# TYPE exposer_request_latencies summary
exposer_request_latencies_count 16166
exposer_request_latencies_sum 10217747.000000
exposer_request_latencies_quantile{quantile="0.500000"} 531.000000
exposer_request_latencies_quantile{quantile="0.900000"} 582.000000
exposer_request_latencies_quantile{quantile="0.990000"} 631.000000
# HELP collector_stats metrics related to collector
# TYPE collector_stats gauge
collector_stats{status="process",type="EdrFlow"} 1166.000000
collector_stats{status="drop",type="EdrFlow"} 319.000000
# HELP collector_stats metrics related to collector
# TYPE collector_stats gauge
collector_stats{status="process",type="EdrHttp"} 12.000000
collector_stats{status="drop",type="EdrHttp"} 0.000000

[exekias]

Thanks for the detailed report. From what I see collector_stats is somehow duplicated and that's causing the client library to fail when parsing the metrics with this error:
text format parsing error in line 19: second HELP line for metric name "collector_stats"

This error is not currently reported but I think we should change that.

[exekias]

this module was refactored here: #9948, I think the problem should be gone now, please reopen if you are able to reproduce again with 7.0 (once released)