Metric & Arrow Schema
Exact schema used when exporting Micrometer metrics as Apache Arrow.
Supported Meter Types
All core Micrometer meters are supported:
- Counter
- Gauge
- Timer
- DistributionSummary
- LongTaskTimer
- FunctionCounter
- FunctionTimer
Arrow Schema (Authoritative)
Each published metric is written as one Arrow row using the following schema:
name STRING NOT NULL
type STRING NOT NULL
application_id STRING
application_name STRING
application_host STRING
tags MAP<STRING, STRING> NOT NULL
value DOUBLE
min DOUBLE
max DOUBLE
mean DOUBLE
This schema is defined in ArrowMetricSchema and is stable across releases.
Field Semantics
| Field | Description |
|---|---|
| name | Metric name |
| type | Meter type (counter, timer, gauge, ...) |
| application_id | Logical application identifier |
| application_name | Human‑readable service name |
| application_host | Hostname or node id |
| tags | Micrometer tags as key‑value map |
| value | Primary metric value (count, gauge value, active tasks) |
| min | Minimum observed value (when applicable) |
| max | Maximum observed value |
| mean | Mean / average value |
Type‑Specific Mapping
| Meter Type | value | min | max | mean |
|---|---|---|---|---|
| Counter | count | 0 | 0 | 0 |
| Gauge | value | 0 | 0 | 0 |
| Timer | count | 0 | max(sec) | mean(sec) |
| DistributionSummary | count | 0 | max | mean |
| LongTaskTimer | active tasks | 0 | total(sec) | avg(sec) |
| Function meters | computed | 0 | computed | computed |
Deterministic Output
The registry guarantees:
- Stable schema
- Deterministic ordering (sorted by type and name)
- Repeatable snapshots
This enables:
- CI regression checks
- Release comparisons
- Diff‑based validation