Selective Test Rerun
Selective Test Rerun governs CI acceleration, shaping system behavior under specific failure modes and workloads.
Definition
Selective Test Rerun is a design property concerning CI acceleration. It defines how the system behaves under specific conditions — what guarantees hold, what trade-offs apply, and how the mechanism interacts with related techniques. The property is only meaningful when its preconditions are met; outside those conditions it provides no benefit and may add cost.
Context and scope
Apply Selective Test Rerun when designing systems where CI acceleration matters. It is most useful under partial failure, high load, or correctness requirements. Pair it with observability to verify the claimed behavior in production. Do not assume the property holds without testing the specific configuration and workload.
Common misconceptions
- Selective Test Rerun eliminates all problems in its domain — it only bounds or shapes specific failure modes, not all pathologies.
- Selective Test Rerun is free once you adopt the right tool — the property depends on configuration, workload, and failure pattern, not the product label alone.
Anti-patterns
- Enabling Selective Test Rerun without measuring its cost in latency, availability, or operational complexity under production-like conditions.
- Assuming Selective Test Rerun composes safely with other guarantees — combining mechanisms can introduce new failure modes or negate benefits.
Related concepts
Authoritative source
What it is
Selective Test Rerun is a design property concerning CI acceleration. It defines how the system behaves under specific conditions — what guarantees hold, what trade-offs apply, and how the mechanism interacts with related techniques. The property is only meaningful when its preconditions are met; outside those conditions it provides no benefit and may add cost.
When to use
Apply Selective Test Rerun when designing systems where CI acceleration matters. It is most useful under partial failure, high load, or correctness requirements. Pair it with observability to verify the claimed behavior in production. Do not assume the property holds without testing the specific configuration and workload.
Common misconceptions
- Selective Test Rerun eliminates all problems in its domain — it only bounds or shapes specific failure modes, not all pathologies.
- Selective Test Rerun is free once you adopt the right tool — the property depends on configuration, workload, and failure pattern, not the product label alone.
Anti-patterns to avoid
- Enabling Selective Test Rerun without measuring its cost in latency, availability, or operational complexity under production-like conditions.
- Assuming Selective Test Rerun composes safely with other guarantees — combining mechanisms can introduce new failure modes or negate benefits.
Related concepts
- Test Impact Analysis (test-impact-analysis)
- Test Parallelization and Sharding (test-parallelization)