feat(edge): add offline read cache for allowlisted State Hub GET routes
All checks were successful
CI Smoke / host-smoke (push) Successful in 0s
CI Smoke / container-smoke (push) Successful in 2s
Build and Publish Multi-Context Image / build-and-push (push) Successful in 51s

The edge relay now persists successful GET responses and serves them with
stale markers when upstream is unreachable. Extend Forgejo image workflow
path filters so api changes trigger registry publishes.
This commit is contained in:
tegwick 2026-07-09 01:04:15 +02:00
parent 138299293d
commit 1cf949bda4
5 changed files with 354 additions and 1 deletions

View file

@ -12,6 +12,9 @@ on:
paths:
- ".forgejo/workflows/image.yaml"
- "Dockerfile"
- "api/**"
- "pyproject.toml"
- "uv.lock"
workflow_dispatch:
env:

152
api/edge/read_cache.py Normal file
View file

@ -0,0 +1,152 @@
from __future__ import annotations
import os
import sqlite3
import stat
from dataclasses import dataclass
from datetime import datetime, timezone
from pathlib import Path
from urllib.parse import urlencode
DEFAULT_READ_CACHE_PATH = Path(
os.environ.get("STATEHUB_READ_CACHE_PATH", "~/.statehub/edge-read-cache.sqlite3")
).expanduser()
MAX_CACHE_ENTRY_BYTES = 1024 * 1024
CACHEABLE_GET_PREFIXES = (
"/state/",
"/workplans/",
"/messages/",
"/decisions/",
"/tasks/",
"/progress/",
"/sbom/",
"/repos",
"/legacy-meter/",
)
@dataclass(frozen=True)
class ReadCacheEntry:
cache_key: str
status_code: int
content: bytes
content_type: str | None
cached_at: str
def default_read_cache_path() -> Path:
return DEFAULT_READ_CACHE_PATH
def utcnow() -> str:
return datetime.now(tz=timezone.utc).isoformat()
def build_cache_key(method: str, path: str, query_items: list[tuple[str, str]]) -> str:
normalized_path = path if path.startswith("/") else f"/{path}"
query = urlencode(sorted(query_items))
if query:
return f"{method.upper()}:{normalized_path}?{query}"
return f"{method.upper()}:{normalized_path}"
def is_cacheable_get(path: str) -> bool:
normalized = path if path.startswith("/") else f"/{path}"
if normalized.startswith("/edge/"):
return False
return any(normalized.startswith(prefix) for prefix in CACHEABLE_GET_PREFIXES)
class ReadCacheStore:
def __init__(self, path: str | Path | None = None) -> None:
self.path = Path(path).expanduser() if path is not None else default_read_cache_path()
self.path.parent.mkdir(parents=True, exist_ok=True)
self._init_db()
self._chmod_private()
def _connect(self) -> sqlite3.Connection:
conn = sqlite3.connect(self.path)
conn.row_factory = sqlite3.Row
return conn
def _init_db(self) -> None:
with self._connect() as conn:
conn.execute(
"""
CREATE TABLE IF NOT EXISTS read_cache_entries (
cache_key TEXT PRIMARY KEY,
status_code INTEGER NOT NULL,
content BLOB NOT NULL,
content_type TEXT,
cached_at TEXT NOT NULL
)
"""
)
conn.execute(
"CREATE INDEX IF NOT EXISTS ix_read_cache_cached_at ON read_cache_entries(cached_at)"
)
conn.commit()
def _chmod_private(self) -> None:
try:
os.chmod(self.path, stat.S_IRUSR | stat.S_IWUSR)
except OSError:
pass
def put(
self,
*,
cache_key: str,
status_code: int,
content: bytes,
content_type: str | None,
) -> None:
if len(content) > MAX_CACHE_ENTRY_BYTES:
return
now = utcnow()
with self._connect() as conn:
conn.execute(
"""
INSERT INTO read_cache_entries (
cache_key, status_code, content, content_type, cached_at
) VALUES (?, ?, ?, ?, ?)
ON CONFLICT(cache_key) DO UPDATE SET
status_code = excluded.status_code,
content = excluded.content,
content_type = excluded.content_type,
cached_at = excluded.cached_at
""",
(cache_key, status_code, content, content_type, now),
)
conn.commit()
def get(self, cache_key: str) -> ReadCacheEntry | None:
with self._connect() as conn:
row = conn.execute(
"SELECT * FROM read_cache_entries WHERE cache_key = ?",
(cache_key,),
).fetchone()
if row is None:
return None
return ReadCacheEntry(
cache_key=row["cache_key"],
status_code=row["status_code"],
content=row["content"],
content_type=row["content_type"],
cached_at=row["cached_at"],
)
def summary(self) -> dict[str, int | str | None]:
with self._connect() as conn:
row = conn.execute(
"""
SELECT COUNT(*) AS entry_count,
MIN(cached_at) AS oldest_cached_at
FROM read_cache_entries
"""
).fetchone()
return {
"entry_count": int(row["entry_count"] or 0),
"oldest_cached_at": row["oldest_cached_at"],
}

View file

@ -2,6 +2,7 @@ from __future__ import annotations
import os
import socket
from datetime import datetime, timezone
from typing import Any
import httpx
@ -9,6 +10,12 @@ from fastapi import FastAPI, Request
from fastapi.responses import JSONResponse, Response
from api.edge.outbox import OutboxEnvelope, OutboxStore, PayloadRejected, default_outbox_path
from api.edge.read_cache import (
ReadCacheStore,
build_cache_key,
default_read_cache_path,
is_cacheable_get,
)
from api.services.write_idempotency import route_class_for
HOP_BY_HOP_HEADERS = {
@ -100,15 +107,35 @@ async def replay_pending(
return counts
def _cache_age_seconds(cached_at: str) -> int:
try:
cached = datetime.fromisoformat(cached_at)
except ValueError:
return 0
if cached.tzinfo is None:
cached = cached.replace(tzinfo=timezone.utc)
return max(0, int((datetime.now(timezone.utc) - cached).total_seconds()))
def _stale_cache_headers(cached_at: str) -> dict[str, str]:
return {
"X-StateHub-Edge-Cache": "stale",
"X-StateHub-Edge-Cache-Age": str(_cache_age_seconds(cached_at)),
}
def create_app(
*,
upstream_url: str | None = None,
outbox_path: str | None = None,
read_cache_path: str | None = None,
timeout: float = 10.0,
) -> FastAPI:
upstream = (upstream_url or os.environ.get("STATEHUB_UPSTREAM_URL") or os.environ.get("API_BASE") or "http://127.0.0.1:8000").rstrip("/")
store_path = outbox_path or default_outbox_path()
cache_path = read_cache_path or str(default_read_cache_path())
store_instance: OutboxStore | None = None
cache_instance: ReadCacheStore | None = None
def get_store() -> OutboxStore:
nonlocal store_instance
@ -116,6 +143,12 @@ def create_app(
store_instance = OutboxStore(store_path)
return store_instance
def get_cache() -> ReadCacheStore:
nonlocal cache_instance
if cache_instance is None:
cache_instance = ReadCacheStore(cache_path)
return cache_instance
app = FastAPI(title="State Hub Edge Relay", version="0.1.0")
@app.get("/edge/health")
@ -134,6 +167,7 @@ def create_app(
"upstream_reachable": reachable,
"upstream_error": error,
"outbox": get_store().summary(),
"read_cache": get_cache().summary(),
}
@app.post("/edge/replay")
@ -156,6 +190,14 @@ def create_app(
if request.headers.get("content-type"):
headers["Content-Type"] = request.headers["content-type"]
cache_key = None
if request.method == "GET" and is_cacheable_get(api_path):
cache_key = build_cache_key(
request.method,
api_path,
list(request.query_params.multi_items()),
)
try:
async with httpx.AsyncClient(base_url=upstream, timeout=timeout) as client:
response = await client.request(
@ -165,13 +207,36 @@ def create_app(
json=body if body is not None else None,
headers=headers,
)
if cache_key is not None and 200 <= response.status_code < 300:
get_cache().put(
cache_key=cache_key,
status_code=response.status_code,
content=response.content,
content_type=response.headers.get("content-type"),
)
response_headers = _safe_response_headers(response.headers)
if cache_key is not None:
response_headers["X-StateHub-Edge-Cache"] = "hit"
return Response(
content=response.content,
status_code=response.status_code,
headers=_safe_response_headers(response.headers),
headers=response_headers,
media_type=response.headers.get("content-type"),
)
except httpx.HTTPError as exc:
if cache_key is not None:
cached = get_cache().get(cache_key)
if cached is not None:
stale_headers = _stale_cache_headers(cached.cached_at)
if cached.content_type:
stale_headers["Content-Type"] = cached.content_type
return Response(
content=cached.content,
status_code=cached.status_code,
headers=stale_headers,
media_type=cached.content_type,
)
route_class = route_class_for(request.method, api_path)
if route_class is None or request.method not in {"POST", "PATCH"}:
return JSONResponse(

View file

@ -84,6 +84,25 @@ or secret-looking JSON fields. Payloads over 64 KiB are rejected.
statehub outbox retry ENVELOPE_ID
statehub outbox cancel ENVELOPE_ID
## Read Cache (Beachhead V1)
The edge relay also keeps a small SQLite read cache for allowlisted `GET`
routes used by activity-core context resolution and daily triage:
- `/state/*`, `/workplans/*`, `/messages/*`, `/decisions/*`, `/tasks/*`,
`/progress/*`, `/sbom/*`, `/repos*`, `/legacy-meter/*`
While upstream is reachable, successful `GET` responses are cached and marked
`X-StateHub-Edge-Cache: hit`. When upstream is unreachable, the relay serves
the last cached body with HTTP 200 and `X-StateHub-Edge-Cache: stale` plus
`X-StateHub-Edge-Cache-Age`.
Default cache path: `STATEHUB_READ_CACHE_PATH`, recommended alongside the
outbox at `~/.statehub/edge-read-cache.sqlite3`.
Uncached reads still return HTTP 503 during outage. Online-only `POST` routes
remain queueable or rejected per the write allowlist above.
## Recovery Checklist
1. Confirm the central State Hub API is reachable.

View file

@ -0,0 +1,114 @@
import httpx
import pytest
from httpx import ASGITransport, AsyncClient
from api.edge.read_cache import ReadCacheStore, build_cache_key, is_cacheable_get
from api.edge.relay import create_app
class FailingAsyncClient:
def __init__(self, *args, **kwargs):
pass
async def __aenter__(self):
return self
async def __aexit__(self, *exc_info):
return False
async def request(self, *args, **kwargs):
raise httpx.ConnectError("upstream down")
async def get(self, *args, **kwargs):
raise httpx.ConnectError("upstream down")
class SuccessAsyncClient:
def __init__(self, *args, **kwargs):
pass
async def __aenter__(self):
return self
async def __aexit__(self, *exc_info):
return False
async def request(self, method, path, **kwargs):
request = httpx.Request(method, f"http://upstream{path}")
return httpx.Response(
200,
json={"path": path, "ok": True},
request=request,
headers={"content-type": "application/json"},
)
def test_build_cache_key_sorts_query_params() -> None:
key = build_cache_key("GET", "/state/summary", [("b", "2"), ("a", "1")])
assert key == "GET:/state/summary?a=1&b=2"
def test_is_cacheable_get_allows_activity_core_reads() -> None:
assert is_cacheable_get("/state/summary")
assert is_cacheable_get("/workplans/index")
assert is_cacheable_get("/tasks/")
assert is_cacheable_get("/edge/health") is False
def test_read_cache_store_round_trip(tmp_path) -> None:
store = ReadCacheStore(tmp_path / "cache.sqlite3")
store.put(
cache_key="GET:/state/summary",
status_code=200,
content=b'{"ok": true}',
content_type="application/json",
)
entry = store.get("GET:/state/summary")
assert entry is not None
assert entry.status_code == 200
assert entry.content == b'{"ok": true}'
@pytest.mark.asyncio
async def test_relay_caches_successful_get_and_serves_stale_on_outage(tmp_path, monkeypatch):
from api.edge import relay
monkeypatch.setattr(relay.httpx, "AsyncClient", SuccessAsyncClient)
cache_path = tmp_path / "read-cache.sqlite3"
app = create_app(
upstream_url="http://upstream",
outbox_path=str(tmp_path / "outbox.sqlite3"),
read_cache_path=str(cache_path),
)
async with AsyncClient(transport=ASGITransport(app=app), base_url="http://edge") as client:
warm = await client.get("/state/summary")
assert warm.status_code == 200
assert warm.headers.get("x-statehub-edge-cache") == "hit"
monkeypatch.setattr(relay.httpx, "AsyncClient", FailingAsyncClient)
async with AsyncClient(transport=ASGITransport(app=app), base_url="http://edge") as client:
stale = await client.get("/state/summary")
assert stale.status_code == 200
assert stale.headers.get("x-statehub-edge-cache") == "stale"
assert stale.json()["ok"] is True
assert int(stale.headers.get("x-statehub-edge-cache-age", "0")) >= 0
@pytest.mark.asyncio
async def test_relay_returns_503_for_uncached_get_when_upstream_unreachable(tmp_path, monkeypatch):
from api.edge import relay
monkeypatch.setattr(relay.httpx, "AsyncClient", FailingAsyncClient)
app = create_app(
upstream_url="http://upstream",
outbox_path=str(tmp_path / "outbox.sqlite3"),
read_cache_path=str(tmp_path / "read-cache.sqlite3"),
)
async with AsyncClient(transport=ASGITransport(app=app), base_url="http://edge") as client:
response = await client.get("/state/summary")
assert response.status_code == 503
assert "not queueable" in response.json()["error"]