Refinery emits metrics about spans received and dropped. Information on the metrics is available here: https://docs.honeycomb.io/manage-data-volume/sample/honeycomb-refinery/monitor/#sampler-metrics
If you have configured Refinery to send metrics to Honeycomb, a helpful Board Template shows what the various metrics mean and how to interpret them.
If the amount of "dropped spans" looks incorrect in each of the sampler metrics, the spans may be getting dropped due to Refinery process restarts or other parts of telemetry pipeline.
You may notice entries in your refinery logs that contain the line and be wondering if this is something you need to investigate further:
"level":"info","msg":"Dropping trace because of sampling","reason":"rules/trace/non-http traces
Usually, you can confirm in your rules.yaml that it's expected that this event was dropped due to not meeting the conditions of the rule in the reason above.
However if you are unsure you may be able to find more information in the corresponding entry opentelemetry-colletor-gateway logs. If you see the following line then the issue is that this data is being dropped as they are logs, not traces.
""msg":"Exporting failed. Dropping data.","kind":"exporter","data_type":"logs"
Currently refinery is only for sampling traces, if you wish to send log data to Honeycomb you shouldn't send it through your Refinery instance.
Learn more about sending logs to Honeycomb here:
https://docs.honeycomb.io/send-data/logs/
To filter out info level information in your refinery logs you can configure the log level to our recommend level 'warn'.