fix(diff): ignore preview env rows so they cannot shadow production (#85)

`GET /applications/{uuid}/envs` does not return one row per key: it merges
the production vars with the PREVIEW ones into one flat array
(`environment_variables->merge(environment_variables_preview)`,
ApplicationsController@envs v4.1.2). The two relations are complements split
on `is_preview`, with a unique index per (key, resource, is_preview) — so the
same key legitimately arrives twice. `fetchEnv` keyed by `key` alone, and
`Object.fromEntries` keeps the LAST, so cast diffed the manifest against
whichever row Coolify happened to serialize last.

Confirmed on prod: REPORTING_ENABLED came back as {value:"true",
is_preview:false} AND {value:"false", is_preview:true}; cast read the "false"
twin and re-proposed a `change` that could never clear.

That is also why #78 looked like a stale read. Both rows are born equal
(Coolify seeds a preview twin), and syncEnv only ever PATCHes the PRODUCTION
row — so the two diverge for exactly the vars updated in place. Five prod
flags flipped false->true re-proposed on every diff forever, while
created-once vars stayed clean because their twins still agreed. Nothing was
stale: cast was reading the other deployment's value. `real_value` tracked
`value` on every row, exactly as the accessor predicts.

cast declares PRODUCTION env and already says so on every write — syncEnv
sends `is_preview: false` on each bulk upsert. This is the read finally
saying the same thing; the asymmetry was the whole bug. Services and
databases map a single set, so this is a no-op for them.

Tests pin the exact prod shape, both serialization orders (the fix is "drop
preview", not "take the first"), a preview-only key, and rows with no
is_preview field at all.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
claude-hdb 2026-07-16 16:37:32 +00:00
parent 61fd72cea9
commit d083f74bec
2 changed files with 110 additions and 3 deletions

View file

@ -875,13 +875,38 @@ export function aliasLive<T extends { name: string }>(
// The diff picks between them per var (see LiveEnvVar, diffEnv); callers that
// only need one flattened string use flattenEnv below.
//
// PREVIEW ROWS ARE DROPPED, and that is load-bearing rather than tidy.
//
// GET /applications/{uuid}/envs does NOT return one row per key. It MERGES two
// parallel sets into one flat array — the production vars and the PREVIEW ones
// (`environment_variables->merge(environment_variables_preview)`,
// ApplicationsController@envs v4.1.2) — and the two relations are complements
// split on `is_preview`, with a unique index per (key, resource, is_preview). So
// the SAME KEY legitimately arrives TWICE, and keying by `key` alone kept
// whichever row Coolify happened to serialize last.
//
// That is #85, and it is why #78 looked like a "stale read": both rows are born
// equal (Coolify seeds a preview twin), and syncEnv below only ever PATCHes the
// PRODUCTION row — so the two diverge for exactly the vars that were updated in
// place. Five prod flags flipped false->true re-proposed as `change` on every
// diff forever, while created-once vars (NODE_ENV, …) stayed clean because their
// twins still agreed. Nothing was stale; cast was reading the other deployment's
// value.
//
// cast declares PRODUCTION env, and already says so on every WRITE — syncEnv
// sends `is_preview: false` on each bulk upsert. This is the read finally saying
// the same thing. A preview var is another deployment's value for the same name:
// not cast's to compare, and not cast's to write. Services and databases have no
// preview relation at all (their controllers map a single set), so this is a
// no-op for them.
//
// A 404 (a resource we just listed no longer having an envs endpoint — not
// expected in practice, but consistent with treating "gone" as "no env vars")
// collapses to {}; anything else (401, 5xx, network) must surface. Swallowing
// it would make a live resource's env look EMPTY, which turns every one of its
// vars into a spurious create in a diff, and into a spurious "missing" in a
// capture.
async function fetchEnv(
export async function fetchEnv(
client: CoolifyClient,
l: Pick<Live, "kind" | "uuid">,
): Promise<Record<string, LiveEnvVar>> {
@ -889,9 +914,16 @@ async function fetchEnv(
const envs = (await client.get(`/${base}/${l.uuid}/envs`).catch((err) => {
if (err instanceof HttpError && err.status === 404) return [];
throw err;
})) as Array<{ key: string; real_value?: string; value: string }>;
})) as Array<{
key: string;
real_value?: string;
value: string;
is_preview?: boolean;
}>;
return Object.fromEntries(
envs.map((e) => [e.key, { value: e.value, realValue: e.real_value }]),
envs
.filter((e) => e.is_preview !== true)
.map((e) => [e.key, { value: e.value, realValue: e.real_value }]),
);
}

View file

@ -2,6 +2,7 @@ import { describe, expect, it, vi } from "vitest";
import {
attachBackup,
attachServiceDomains,
fetchEnv,
fetchLive,
renderAbsentTarget,
} from "../src/cli.js";
@ -276,6 +277,80 @@ describe("attachBackup", () => {
});
});
describe("fetchEnv — preview rows never shadow production (cast#85)", () => {
const app = { kind: "application" as const, uuid: "a1" };
const client = (body: unknown) =>
new CoolifyClient(
"https://coolify.test",
"tok",
vi.fn(
async () => new Response(JSON.stringify(body), { status: 200 }),
) as unknown as typeof fetch,
);
// The exact shape prod returned (#85): GET /applications/{uuid}/envs merges
// environment_variables with environment_variables_preview, so a key arrives
// TWICE — production first, its preview twin second. Keying by `key` alone
// kept the LAST, which is how five flipped prod flags re-proposed forever.
it("takes the production row even though the preview twin is serialized last", async () => {
const env = await fetchEnv(
client([
{
key: "REPORTING_ENABLED",
value: "true",
real_value: "true",
is_preview: false,
},
{
key: "REPORTING_ENABLED",
value: "false",
real_value: "false",
is_preview: true,
},
]),
app,
);
expect(env.REPORTING_ENABLED).toEqual({ value: "true", realValue: "true" });
});
// Order must not decide the answer: the fix is "drop preview", not "take the
// first" — a serialization order that put the twin first would otherwise just
// move the bug rather than remove it.
it("takes the production row even when the preview twin is serialized FIRST", async () => {
const env = await fetchEnv(
client([
{ key: "BRAIN_ENABLED", value: "false", is_preview: true },
{ key: "BRAIN_ENABLED", value: "true", is_preview: false },
]),
app,
);
expect(env.BRAIN_ENABLED).toEqual({ value: "true", realValue: undefined });
});
// A preview-only var is not production state at all. It must not surface as an
// orphan/remove-candidate either — cast neither compares nor writes it.
it("drops a key that exists ONLY as a preview var", async () => {
const env = await fetchEnv(
client([
{ key: "PORT", value: "3000", is_preview: false },
{ key: "PREVIEW_ONLY", value: "x", is_preview: true },
]),
app,
);
expect(Object.keys(env)).toEqual(["PORT"]);
});
// Services and databases map a single set and their rows may carry no
// is_preview at all — absent must mean "keep", never "drop".
it("keeps rows with no is_preview field (services/databases)", async () => {
const env = await fetchEnv(client([{ key: "APP_SECRET", value: "s" }]), {
kind: "service",
uuid: "s1",
});
expect(env.APP_SECRET).toEqual({ value: "s", realValue: undefined });
});
});
describe("attachServiceDomains (cast#72)", () => {
const svc = () => ({
kind: "service" as const,