Central was serving two revisions behind the code it shipped: review_contracts did not exist there although its migration was inside the running image. There was no migration mechanism at all — bare uvicorn CMD, nothing chart-declared — and nothing surfaced the mismatch. The API starts happily against a schema it was not built for and only fails when a request touches a missing table. Adds a chart-managed Helm pre-install/pre-upgrade hook running alembic upgrade head, weighted to complete before the API rolls. A hook rather than an init container: init containers run per pod, so more than one replica means concurrent alembic upgrade with no locking. Failed jobs are deliberately retained — a migration that fails and vanishes is how this drifted in the first place. /state/health now reports applied and expected revisions. "unknown" is deliberately not "ok": an instance that cannot establish agreement must not claim it, the same principle as instance_role defaulting to unknown. Refs STATE-WP-0083-T07 Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Assistant: claude-code Assistant-Model: opus Assistant-Process: 2583210@bnt-lap001 Assistant-Session: f2bff2d5-e9b2-4338-92ca-10282a927006
56 lines
1.8 KiB
Python
56 lines
1.8 KiB
Python
"""Schema/code agreement must be observable (STATE-WP-0083-T07).
|
|
|
|
Central served two revisions behind the code with no signal at all. The API
|
|
started fine and would only have failed when a request touched a missing table.
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
import pytest
|
|
|
|
from api.services import schema_state as ss
|
|
|
|
|
|
class _Session:
|
|
def __init__(self, revision=None, raises=False):
|
|
self._revision = revision
|
|
self._raises = raises
|
|
|
|
async def execute(self, *_a, **_k):
|
|
if self._raises:
|
|
raise RuntimeError("no connection")
|
|
rev = self._revision
|
|
|
|
class R:
|
|
def scalar(self_inner):
|
|
return rev
|
|
|
|
return R()
|
|
|
|
|
|
def test_code_head_is_readable_from_the_shipped_migrations():
|
|
"""Must not require a database: it is what the code expects, not what ran."""
|
|
assert ss.code_head_revision()
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_matching_revision_reports_ok(monkeypatch):
|
|
monkeypatch.setattr(ss, "code_head_revision", lambda: "abc123")
|
|
assert (await ss.schema_state(_Session("abc123")))["status"] == "ok"
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_behind_revision_is_reported_with_both_values(monkeypatch):
|
|
monkeypatch.setattr(ss, "code_head_revision", lambda: "newrev")
|
|
state = await ss.schema_state(_Session("oldrev"))
|
|
assert state["status"] == "behind"
|
|
assert state["applied"] == "oldrev" and state["expected"] == "newrev"
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_unknown_is_not_reported_as_ok(monkeypatch):
|
|
"""An instance that cannot establish agreement must not claim it."""
|
|
monkeypatch.setattr(ss, "code_head_revision", lambda: "newrev")
|
|
assert (await ss.schema_state(_Session(raises=True)))["status"] == "unknown"
|
|
monkeypatch.setattr(ss, "code_head_revision", lambda: None)
|
|
assert (await ss.schema_state(_Session("oldrev")))["status"] == "unknown"
|