Did you run Metribeat in debug mode? (./metricbeat -e -d "*") please only add the specific mbean in the config so as to have less noise and check if logs can give us anything useful.
With your advice we could trace the problem further. Apart from deactivating the unrelevant mbeans, we also deactivated other modules like 'system'. After that we could clearly see that the data is sent without an issue.
Checking the Logs of Logstash we figured that it was not able to ingest the data due to an error in an elasticsearch pipeline.
I wrote this pipeline a while ago to extract useful information from the mbean field. Sadly this field did not exist in the jmx metrics JVM memory. I changed the pipeline to ignore the absence of the field mbean and it started working for a while.
I found another issue but did not analyze it further. We are only interested in the attribute HeapMemoryUsage. Configuring it properly did not yield any data. It only starts working after configuring the attribute NonHeapMemoryUsage additionally.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.