influxdata / influxdata/datafusion-udf-wasm
Metrics
- Dominant language
- Rust
- Stars
- 20
- Forks
- 3
- Avg merge
- 6h 20m
- Merged PRs (30d)
- 24
Description
# What
Hook up [DataFusion metrics](https://docs.rs/datafusion/latest/datafusion/physical_plan/struct.Metric.html).
# Why
Users probably want insights into how their UDFs are performing.
# How
I'm not entirely sure if DataFusion has a good story of per-UDF metrics. It seems that metrics only exists for physical plans, but UDFs are created beforehand (or at least they exist both in the logical and physical realm). So this might need some upstream work.
Some example metrics:
- HTTP requests:
- request count
- number of uploaded bytes
- number of downloaded bytes
- number of epochs consumed by UDF
- number of max bytes (this is probably already exposed by the DF memory pool)
Contributor guide
Research direction
Start by reading DataFusion's physical-plan Metric documentation and tracing how UDFs exist across logical and physical plans. Determine whether per-UDF metrics can be exposed for request counts, transfer bytes, epochs, and memory, or whether upstream DataFusion work is required. Done means a feasible implementation scope is established or the upstream blocker is documented.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- rust, wasm
- Domain
- observability
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100