apache / apache/datafusion-ballista
Make prometheus metrics configurable
- Dominant language
- Rust
- Stars
- 2.1k
- Forks
- 320
- Avg merge
- 1d 22h
- Merged PRs (30d)
- 66
Description
**Is your feature request related to a problem or challenge? Please describe what you are trying to do.**
A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]
(This section helps Arrow developers understand the context and *why* for this feature, in addition to the *what*)
In https://github.com/apache/arrow-ballista/pull/511 we've added prometheus metrics in the default build of the ballista scheduler but the metrics exposed (and more importantly the histogram buckets used) are currently hard-coded.
**Describe the solution you'd like**
A clear and concise description of what you want to happen.
Allow users to configure metrics. This should be along two dimensions:
1. We may support an increasing set of metrics which we can track, but users may only want to track a subset of them. In that case they should be able to toggle on/off metrics.
2. For histogram metrics, they should be able to configure the histogram buckets.
**Describe alternatives you've considered**
A clear and concise description of any alternative solutions or features you've considered.
Users can always implement a custom `SchedulerMetricsCollector` and do whatever they want, but it is pain since they will not be able to use the standard image and will have to build their project themselves and manage their own entrypoint.
Alternatively, we could somehow leverage a plugin mechanism for this
**Additional context**
Add any other context or screenshots about the feature request here.
Contributor guide
Research direction
Start by reviewing pull request 511 and the SchedulerMetricsCollector mentioned in the issue to understand the current hard-coded metrics and histogram buckets. Define how users will select metrics and configure buckets without needing a custom build, then verify that the standard Ballista image supports both forms of configuration.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- prometheus, rust
- Domain
- observability
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 25/100