[Bug] Prometheus metrics plugin will throw Incorrect number of labels when partitioned table and non-partitioned table mixed in cluster

[luoyuxia]

Search before asking

• I searched in the issues and found nothing similar.

Fluss version

main

Minimal reproduce step

1: enable prometheus metric reporter
2: create a partitioned table and non-partitioned table
3: then the following waning message will be thrown

2024-12-27 16:48:12,199 WARN  com.alibaba.fluss.metrics.registry.MetricRegistryImpl        [] - Error while handling metric: messagesInPerSecond.
java.lang.IllegalArgumentException: Incorrect number of labels.
	at io.prometheus.client.SimpleCollector.setChild(SimpleCollector.java:136) ~[?:?]
	at com.alibaba.fluss.metrics.prometheus.PrometheusReporter.addMetric(PrometheusReporter.java:214) ~[?:?]
	at com.alibaba.fluss.metrics.prometheus.PrometheusReporter.notifyOfAddedMetric(PrometheusReporter.java:153) ~[?:?]
	at com.alibaba.fluss.metrics.registry.MetricRegistryImpl.forAllReporters(MetricRegistryImpl.java:267) ~[fluss-server-0.6-SNAPSHOT.jar:0.6-SNAPSHOT]
	at com.alibaba.fluss.metrics.registry.MetricRegistryImpl.register(MetricRegistryImpl.java:214) ~[fluss-server-0.6-SNAPSHOT.jar:0.6-SNAPSHOT]
	at com.alibaba.fluss.metrics.groups.AbstractMetricGroup.addMetric(AbstractMetricGroup.java:292) ~[fluss-server-0.6-SNAPSHOT.jar:0.6-SNAPSHOT]
	at com.alibaba.fluss.metrics.groups.AbstractMetricGroup.meter(AbstractMetricGroup.java:250) ~[fluss-server-0.6-SNAPSHOT.jar:0.6-SNAPSHOT]

What doesn't meet your expectations?

Should print warn message

Anything else?

1: for partitioned table and metrics name messagesInPerSecond, assuming m1, it'll be of scope name fluss_tabletserver_table__messagesInPerSecond with labels {database=test_db_1, cluster_id=fluss, partition=2027, host=127.0.0.1, tablet_type=kv, server_id=2, table=test_pk_table_1}. Note the label count is 7

2: for non-partitioned table and metrics name fluss_tabletserver_table__messagesInPerSecond, assuming m2, it'll be of scope name
fluss_tabletserver_table__messagesInPerSecond with labels {database=test_db_11, cluster_id=fluss, host=127.0.0.1, tablet_type=kv, server_id=2, table=tt}. Note the label count is 6

First, it registers m1 with a new created Collector c1 to Prometheus reporter. Then when add m2, it has same scope name with m1, so it use c1 to add. However, c1 is expected to with label count 7, m2 is with label count 6. So, Incorrect number of labels. is thrown..

Are you willing to submit a PR?

• I'm willing to submit a PR!