Chart for per-SP Spark Retrieval Result Codes breakdown

[juliangruber]

Based on this feedback:

It would be great if the spark platform would open up the retrieval logs, we can see the official retrieval logs at the moment, but the frequency is very low, resulting in a very low retrieval rate for us, and this issue is confusing us!

Source: https://space-meridian.slack.com/archives/C02T827T9N0/p1730970300183279?thread_ts=1730734156.407589&cid=C02T827T9N0

I was thinking just about this today, when running the fetch-recent-miner-measurements.js script. Can we expose that information through an API?

I think this would best be a chart:

This, but per SP

[juliangruber]

Currently, we're recording retrieval stats at a network level in InfluxDB:

https://github.com/filecoin-station/spark-evaluate/blob/2d5608acbac88eb7bd190bd1e9ca91172ffc530d/lib/evaluate.js#L153-L156

This is then observed, persisted and re-exposed in spark-stats:

https://github.com/filecoin-station/spark-stats/blob/ba50a07bff3f61e476e92c6feead99c7dac6a2a6/observer/lib/observer.js#L100-L131

and

https://github.com/filecoin-station/spark-stats/blob/ba50a07bff3f61e476e92c6feead99c7dac6a2a6/stats/lib/stats-fetchers.js#L254-L272

We could either add a SP dimension to the data in InfluxDB, or move this to spark-evaluate's Postgres database.