feat: query path instrumentation

[praveen-influx]

• spans added for buffer, parquet chunks along with number of files that are already in parquet cache along with the sql

NOTE: The parquet cache check for files is done at the point of call to TableProvider::scan, the data may get evicted by the time datafusion actually tries to access it but for a lot of the queries which are running in perf test env, this is a good enough approximation for now

The workflow here is,

• spin up influxdb3 with extra switches --traces-exporter-jaeger-agent-port 6831 --traces-exporter-jaeger-service-name influxdb3-core --traces-exporter jaeger and
• jaeger - following docs

Once it's running, issue queries and then in jaeger ui (running on http://localhost:16686), search for

• service name: influxdb3-core
• operation: query_database_sql
• tags: query="select count(*) from cpu_4 where time > now() - interval '15 minutes' limit 1"

That should return bunch of traces and drilling into one of those get_db_namespace should show table_chunks span that shows the buffer chunks, parquet chunks and how many parquet chunks are already in the cache

[waynr]

👍