While attempting to configure a Prometheus exporter I discovered that counter metrics were not behaving as expected, their values would fluctuate up and down and the metrics would disappear from the scrape endpoint when the system was not active. I believe this is because prometheus.NewExportPipeline() missed the change to a stateful batcher from #385 when merge conflicts were resolved in #395.
Thanks for reporting this right @Aneurysm9. You're right, the batcher should be stateful things must be messed with these merges :)
I've filled #428 to fix that. Once that is merged, could you check if this fixes the problem you're reporting? Otherwise we may need some more information to debug it.
Addressed in v0.2.1.1
@paivagustavo thanks for the quick fix. I can confirm that the exporter is now exhibiting the expected behavior.
Happy that it was fixed, thank you for getting back!
Most helpful comment
@paivagustavo thanks for the quick fix. I can confirm that the exporter is now exhibiting the expected behavior.