Redress
redress (v.): to remedy or to set right.
redress is a failure-policy library for Python services.
It treats retries, circuit breakers, and stop conditions as coordinated responses to classified failure, rather than independent wrappers. The goal is to make failure behavior explicit, bounded, and observable across a codebase.
Why redress?
- Per-class backoff: Tune retries by coarse error class (429 vs. 5xx vs. deadlocks).
- Pluggable classifiers: Built-ins for HTTP status + SQLSTATE, plus optional extras for aiohttp/grpc/boto3/redis/urllib3.
- Sync/async symmetry: Same semantics for threads and asyncio, plus a zero-arg
@retrydecorator. - Deterministic envelopes: Deadlines, max attempts, and caps for unknown errors.
- Best-effort observability: Metric/log hooks that never break workloads.
Quick start
from redress import retry
@retry # default_classifier + decorrelated_jitter(max_s=5.0)
def fetch_user():
...
Async is the same:
@retry
async def fetch_user_async():
...
Prefer policies directly?
from redress import Policy, Retry, default_classifier
from redress.strategies import decorrelated_jitter
policy = Policy(
retry=Retry(
classifier=default_classifier,
strategy=decorrelated_jitter(max_s=5.0),
deadline_s=30,
max_attempts=5,
)
)
result = policy.call(lambda: do_work(), operation="sync_task")
What’s inside
- API highlights:
Policy/Retry,CircuitBreaker,RetryPolicy/AsyncRetryPolicy,@retry, classifiers (default,http_classifier,sqlstate_classifier,aiohttp_classifier,grpc_classifier,boto3_classifier,redis_classifier,urllib3_classifier,pyodbc_classifier), strategies (decorrelated_jitter,equal_jitter,token_backoff), hooks (on_metric,on_log), context manager reuse. - Use cases: HTTP 429/5xx, DB deadlocks/SQLSTATE 40001, queue/worker retries, third-party API calls, async services.
- Production pointers: Set
deadline_sandmax_attempts, capmax_unknown_attempts, keep tags low-cardinality (class,operation,err), attach metrics/log hooks.
Where to go next
- Getting started – install and first examples.
- Concepts – error classes, strategies, policies, decorators.
- Observability – metrics/log hooks and patterns.
- Safety and resilience: hook isolation, jitter guidance, production checklist.
- Usage - basic usage patterns.
- Recipes - HTTP/DB/worker patterns and observability wiring.
- Testing utilities – deterministic helpers and policy spies.
- Reference flows - end-to-end service patterns.
- Examples & Integrations – runnable snippets for HTTP, DB, workers, FastAPI, benchmarks.
- Comparison – Redress vs. alternatives.
- API reference – entry points at a glance.