'Flink taskmanager metrics disappear after starting a job

I have a Standalone cluster (with jobmanager and taskmanager on same machine) on 1.14.4 and I'm testing the migration to 1.15.0

But I keep losing the taskmanager metrics when I start a job on the 1.15 cluster

I use the same configuration as in the previous cluster

  metrics.reporters: prom
  metrics.reporter.prom.factory.class: org.apache.flink.metrics.prometheus.PrometheusReporterFactory
  metrics.reporter.prom.port: 9250-9251

If the cluster is running without jobs I can see the metrics on port 9250 for jobmanager and on port 9251 for taskmanager

If I start a job, the metrics from taskmanager disappear and if I stop the job the metrics come live again

What am I missing?

Response

Its a bug on metrics of kafka source and sink on 1.15.0 https://lists.apache.org/thread/6bd9vmcroh7576d7h1kdcd8czf0b4l73

Is necessary to disable the registration of metrics on both

For source:

kafkaSourceBuilder.setProperty("register.consumer.metrics", "false");

For sink:

var producerProperties = new Properties();
producerProperties.setProperty("register.producer.metrics", "false");
producerSinkBuilder.setKafkaProducerConfig(producerProperties);


Sources

This article follows the attribution requirements of Stack Overflow and is licensed under CC BY-SA 3.0.

Source: Stack Overflow

Solution Source