Issue with histogram counters being stuck

[Pascal Sick]

Hello Group,

We are using MicroProfile (6.1) @Timed with a REST API implementation and I'm seeing weird counters in histogram values.

le="+Inf" - everything is fine, it counts up continuously (and resets correctly)
le="*" - anything else gets stuck at 1028

Config:

mp.metrics.distribution.timer.buckets=.*=20ms,50ms,100ms,300ms,700ms,1000ms,1500ms,3s,5s,10s,30s

Example:
..._seconds_bucket{mp_scope="application",le="1.0"} 1028.0

... all up and incl. le="30.0" have value 1028
..._seconds_bucket{mp_scope="application",le="+Inf"} 308270.0

I tried to find anything on the web, but without luck, so I'm turning to this group for hints on what we did wrong or which buttons to push to fix this. Unless, ofc, we're the first ones to stumble over a bug that causes this, but this seems unlikely.

Thank you!
Pascal

[Emily Jiang]

Hi Pascal,

Looks like you are using MicroProfile Metrics. Please create an issue on that spec for a detailed discussion.

By the way, MicroProfile Metrics was deprecated and the successor is MicroProfile Telemetry, based on OpenTelemetry. You might want to adopt MicroProfile Telemetry for ongoing development.

Thanks,

Emily

[Pascal Sick]

Hi Emily,

thanks a lot for the feedback! We will definitely look into it again after updating to MicroProfileTelemetry.

Best regards

Pascal