fix(org): clean up every site when deleting an org

deleteOrgById fetched the org's sites with .limit(1), so the per-site
cleanup loop ran for at most one site: the WireGuard deletePeer call,
the Newt terminate collection, and cleanupSiteAssociations were all
skipped for every other site in the org.

Drop the limit so every site is loaded, and mirror deleteSite.ts in the
loop: run cleanupSiteAssociations for newt sites so clients of other
orgs do not keep stale associations for the deleted sites.

Covered by a standalone regression test that seeds two wireguard and
two newt sites plus a stale association-cache row, runs deleteOrgById
against a stubbed exit node, and asserts all four sites were cleaned.
On the previous code the test fails with exactly the reported symptom:
one peer removal instead of two.

Fixes #3863
This commit is contained in:
choksi2212
2026-10-13 06:52:56 +05:30
parent 6bab69dccb
commit dc579acc44
3 changed files with 343 additions and 15 deletions
+185
View File
@@ -0,0 +1,185 @@
import { execFileSync } from "child_process";
import fs from "fs";
import os from "os";
import path from "path";
import { fileURLToPath } from "url";
import { assertEquals } from "@test/assert";
const here = path.dirname(fileURLToPath(import.meta.url));
const repoRoot = path.resolve(here, "..", "..");
const CHILD_SCRIPT_PATH = path.join(
repoRoot,
"server",
"lib",
"deleteOrg.childTest.ts"
);
function runChild(workdir: string): {
exitCode: number;
observations: any;
output: string;
} {
const tsconfig = ["tsconfig.json", "tsconfig.oss.json"]
.map((file) => path.join(repoRoot, file))
.find((file) => fs.existsSync(file));
if (!tsconfig) {
throw new Error("No tsconfig found for @server path aliases");
}
const tsxCli = path.join(
repoRoot,
"node_modules",
"tsx",
"dist",
"cli.mjs"
);
if (!fs.existsSync(tsxCli)) {
throw new Error("tsx is not installed; run npm ci first");
}
try {
const output = execFileSync(
process.execPath,
[tsxCli, "--tsconfig", tsconfig, CHILD_SCRIPT_PATH],
{
cwd: workdir,
timeout: 120000,
encoding: "utf8",
env: { ...process.env, NODE_ENV: "test" }
}
);
const m = output.match(/OBSERVATIONS (.*)/);
if (!m) {
throw new Error("child did not print observations: " + output);
}
return { exitCode: 0, observations: JSON.parse(m[1]), output };
} catch (error: any) {
const output = String(error?.stdout ?? "");
return { exitCode: 1, observations: null, output };
}
}
function createTestEnvironment(): string {
for (const generated of ["server/build.ts", "server/db/index.ts"]) {
if (!fs.existsSync(path.join(repoRoot, generated))) {
throw new Error(
`Missing ${generated}; run npm run set:oss && npm run set:sqlite first`
);
}
}
const workdir = fs.mkdtempSync(
path.join(os.tmpdir(), "pangolin-org-cleanup-")
);
fs.mkdirSync(path.join(workdir, "config", "db"), { recursive: true });
fs.copyFileSync(
path.join(repoRoot, "config", "config.example.yml"),
path.join(workdir, "config", "config.yml")
);
fs.symlinkSync(
path.join(repoRoot, "server"),
path.join(workdir, "server"),
process.platform === "win32" ? "junction" : "dir"
);
fs.symlinkSync(
path.join(repoRoot, "test"),
path.join(workdir, "test"),
process.platform === "win32" ? "junction" : "dir"
);
fs.symlinkSync(
path.join(repoRoot, "node_modules"),
path.join(workdir, "node_modules"),
process.platform === "win32" ? "junction" : "dir"
);
return workdir;
}
function applyMigrations(workdir: string) {
const tsconfig = ["tsconfig.json", "tsconfig.oss.json"]
.map((file) => path.join(repoRoot, file))
.find((file) => fs.existsSync(file));
if (!tsconfig) {
throw new Error("No tsconfig found for @server path aliases");
}
const tsxCli = path.join(
repoRoot,
"node_modules",
"tsx",
"dist",
"cli.mjs"
);
// CI applies migrations with `npm run db:push` before booting the app;
// server/db/migrate.ts is the drizzle migrator that script invokes.
const migrationsScript = path.join(repoRoot, "server", "db", "migrate.ts");
execFileSync(
process.execPath,
[tsxCli, "--tsconfig", tsconfig, migrationsScript],
{
cwd: workdir,
timeout: 120000,
encoding: "utf8",
env: { ...process.env, NODE_ENV: "test" }
}
);
}
function testOrgCleanupCoversEverySite() {
console.log("Running deleteOrgById per-site cleanup test...");
const workdir = createTestEnvironment();
applyMigrations(workdir);
const { exitCode, observations, output } = runChild(workdir);
if (exitCode !== 0) {
throw new Error("deleteOrgById child run failed: " + output);
}
// The exact bug from the issue: with a single-row sites fetch, only one
// of the four sites gets its side effects. All wireguard peers must be
// removed from the exit node...
assertEquals(
JSON.stringify(observations.peerRemovals),
JSON.stringify(["wgKey1", "wgKey2"]),
"deleteOrgById must remove every wireguard site's peer from its exit node"
);
// ...and every newt connector must be collected for the terminate message.
assertEquals(
JSON.stringify(observations.terminatedNewtIds),
JSON.stringify(["newtId3", "newtId4"]),
"deleteOrgById must collect every newt id so all connectors get the terminate message"
);
// DB rows must still cascade to nothing left.
assertEquals(observations.sitesLeft, 0, "all site rows must be gone");
assertEquals(observations.newtsLeft, 0, "all newt rows must be gone");
assertEquals(
observations.sessionsLeft,
0,
"all newt session rows must be gone"
);
// The gap the issue calls out separately: cleanupSiteAssociations must
// run for newt sites (deleteSite.ts already does this), so the stale
// association cache rows for these sites must be cleared.
assertEquals(
observations.associationRowsLeft,
0,
"newt site cleanup must clear clientSitesAssociationsCache rows"
);
console.log("deleteOrgById per-site cleanup test passed");
}
function main() {
let failed = 0;
for (const test of [testOrgCleanupCoversEverySite]) {
try {
test();
} catch (error) {
failed++;
console.error(`FAIL: ${(error as Error).message}`);
}
}
if (failed > 0) {
process.exit(1);
}
console.log("deleteOrg tests complete");
}
main();