I’m trying to replace our old Java exporter with this new one, but the metrics seem to behave differently.
With the old exporter, the query shows lag increasing gradually over time when there’s no consumption from a topic.
However, the new Klag exporter reports a sharp spike for the same metric, which doesn’t seem correct—if there’s lag, it should grow incrementally as time passes.
Could you please take a look?
metric:
max by (group, topic, partition) (kafka_consumergroup_group_lag_seconds{cluster_name=~"$cluster",app_kubernetes_io_instance="kafka-lag-exporter", group=~"$consumer_group", topic=~"$topic"})
old behavior:
new exporter behavior:
I’m trying to replace our old Java exporter with this new one, but the metrics seem to behave differently.
With the old exporter, the query shows lag increasing gradually over time when there’s no consumption from a topic.
However, the new Klag exporter reports a sharp spike for the same metric, which doesn’t seem correct—if there’s lag, it should grow incrementally as time passes.
Could you please take a look?
metric:
old behavior:
new exporter behavior: