-
Enhancement
-
Resolution: Not a Bug
-
Major
-
None
-
None
-
False
-
-
False
We have configured Debezium using the debezium operator.
We have enabled the JMX exporter by setting spec.runtime.metrics.jmxExporter.enabled to True.
The resulting metrics are of the form:
$ curl -s http://localhost:9090/metrics | grep ^debezium_metrics | sed 's:.*{\(.*\)}.*:\1:' | sort | uniq -c 19 context="snapshot",name="table",plugin="postgres" 14 context="streaming",name="table",plugin="postgres" $ curl -s http://localhost:9090/metrics | grep debezium_metrics_TotalNumberOfUpdateEventsSeen # HELP debezium_metrics_TotalNumberOfUpdateEventsSeen debezium.postgres:name=null,type=connector-metrics,attribute=TotalNumberOfUpdateEventsSeen # TYPE debezium_metrics_TotalNumberOfUpdateEventsSeen gauge debezium_metrics_TotalNumberOfUpdateEventsSeen{context="snapshot",name="table",plugin="postgres"} 0.0 debezium_metrics_TotalNumberOfUpdateEventsSeen{context="streaming",name="table",plugin="postgres"} 2515.0 $
The value of the name label seems to be the value of the topic prefix, which is the same for all tables handled by the particular Debezium server.
Which use case/requirement will be addressed by the proposed feature?
It would be nice to be able to break down the value of metrics per table for monitoring, debezium_metrics_TotalNumberOfUpdateEventsSeen for example.
Debezium image: quay.io/debezium/server:3.0.7.Final
