Implement Phase 0b base project against the frozen test contract

All seven services (gateway, ledger, fraud, sanctions, insights, notifier,
audit), db/schema.sql plus optional_indexes.sql, scripts/seed.py and
smoke.py, the gateway-served frontend, and docker-compose.yml. Plain-text
stdlib logging and a weak manual X-Request-Id only -- no observability yet,
by design, per test_no_observability.py. Implements all 13 named chaos-flag
pathologies from docs/PATHOLOGIES.md behind Redis-backed runtime flags.

make smoke and make test-contract pass; make test-baseline and
make test-pathology pass except for a handful of load-sensitive tests
documented in tests/README.md rather than papered over.
This commit is contained in:
Movses Aslanyan
2026-09-05 08:12:30 +03:30
parent ccf1783779
commit 91afea92a1
51 changed files with 2757 additions and 0 deletions
View File
+75
View File
@@ -0,0 +1,75 @@
"""Shared low-level access to the `chaos:flags` Redis hash.
This is infrastructure (a Redis hash reader/writer), not pathology behaviour --
the mechanism each flag triggers lives in that service's own chaos.py, per
phase-0-base-project-spec.md §7 ("each pathology lives in a clearly named
chaos.py module inside its owning service"). Every service re-reads the hash
on every request; nothing here is cached in-process.
"""
import random
import time
from .redis_client import redis_client
FLAGS_KEY = "chaos:flags"
ALL_FLAGS = [
"slow_sanctions",
"sanctions_errors",
"n_plus_one",
"pool_starvation",
"row_lock_contention",
"blocking_event_loop",
"retry_storm",
"no_batching",
"unbounded_cache",
"slow_consumer",
"cold_cache",
"poison_data",
"chatty_health",
]
NAMED_PROFILES = {
"clean": set(),
"light": {"slow_sanctions", "cold_cache"},
"nightmare": set(ALL_FLAGS),
}
async def get_flags() -> dict:
raw = await redis_client.hgetall(FLAGS_KEY)
return {flag: raw.get(flag) == "1" for flag in ALL_FLAGS}
async def is_enabled(flag: str) -> bool:
return await redis_client.hget(FLAGS_KEY, flag) == "1"
async def set_flag(flag: str, enabled: bool) -> None:
if flag not in ALL_FLAGS:
raise ValueError(f"unknown flag {flag!r}")
await redis_client.hset(FLAGS_KEY, flag, "1" if enabled else "0")
async def set_profile(profile: str) -> dict:
if profile == "random":
# Seed derived from time.time(), never logged -- the active set lives
# only in the `chaos:flags` hash, per §7's "random" profile note.
rng = random.Random(time.time())
chosen = set(rng.sample(ALL_FLAGS, k=rng.randint(3, 6)))
elif profile in NAMED_PROFILES:
chosen = NAMED_PROFILES[profile]
else:
raise ValueError(f"unknown profile {profile!r}")
mapping = {flag: ("1" if flag in chosen else "0") for flag in ALL_FLAGS}
await redis_client.hset(FLAGS_KEY, mapping=mapping)
return await get_flags()
async def flags_fingerprint() -> str:
"""A short, stable string identifying the current flag combination. Used
to key the insights summary cache so a live flag change (G5) is never
masked by a stale cache entry written under a different chaos state."""
flags = await get_flags()
return "".join("1" if flags[f] else "0" for f in ALL_FLAGS)
+15
View File
@@ -0,0 +1,15 @@
from sqlalchemy.ext.asyncio import async_sessionmaker, create_async_engine
from .settings import settings
def make_engine(**overrides):
"""A fresh async engine. Services that need more than one pool profile
(see ledger/chaos.py for pool_starvation) call this more than once."""
kwargs = dict(pool_pre_ping=True)
kwargs.update(overrides)
return create_async_engine(settings.postgres_dsn, **kwargs)
engine = make_engine(pool_size=20, max_overflow=10)
Session = async_sessionmaker(engine, expire_on_commit=False)
+13
View File
@@ -0,0 +1,13 @@
import httpx
from .settings import settings
def downstream_client(request_id: str) -> httpx.AsyncClient:
"""An httpx.AsyncClient preconfigured with the 3s downstream timeout and
the X-Request-Id header forwarded, per phase-0-base-project-spec.md §9 --
this header is the only correlation mechanism that exists in Phase 0."""
return httpx.AsyncClient(
timeout=settings.downstream_timeout_seconds,
headers={"X-Request-Id": request_id},
)
+12
View File
@@ -0,0 +1,12 @@
import logging
def configure_logging(service_name: str) -> logging.Logger:
"""Plain-text, human-prose logging -- the deliberately weak Phase 0
baseline. Per phase-0-base-project-spec.md §9: no JSON, no key-value, no
structured fields. Do not "improve" this; later phases replace it."""
logging.basicConfig(
level=logging.INFO,
format="%(asctime)s %(levelname)s %(name)s %(message)s",
)
return logging.getLogger(service_name)
+5
View File
@@ -0,0 +1,5 @@
import redis.asyncio as redis
from .settings import settings
redis_client = redis.from_url(settings.redis_url, decode_responses=True)
+27
View File
@@ -0,0 +1,27 @@
from pydantic_settings import BaseSettings, SettingsConfigDict
class Settings(BaseSettings):
model_config = SettingsConfigDict(env_file=".env", extra="ignore")
service_name: str = "unknown"
postgres_dsn: str = "postgresql+asyncpg://ledgerline:ledgerline@postgres:5432/ledgerline"
redis_url: str = "redis://redis:6379/0"
gateway_url: str = "http://gateway:8000"
ledger_url: str = "http://ledger:8001"
fraud_url: str = "http://fraud:8002"
sanctions_url: str = "http://sanctions:8003"
insights_url: str = "http://insights:8004"
notifier_url: str = "http://notifier:8005"
downstream_timeout_seconds: float = 3.0
stream_name: str = "stream:transactions"
summary_cache_ttl_seconds: int = 900
summary_cache_ttl_seconds_cold: int = 60
settings = Settings()