Add Plutonium version checks and scheduled rolling restarts

This commit is contained in:
2026-09-21 18:47:24 -04:00
parent 424d6cefb6
commit 47d6dc07b7
10 changed files with 583 additions and 0 deletions
+55
View File
@@ -0,0 +1,55 @@
export interface MaintenanceSchedule {
enabled: boolean;
time: string;
nextAt: string | null;
}
export interface PlutoniumRelease {
revision: number | null;
checkedAt: string | null;
error: string | null;
}
export interface PlutoniumVersion {
server: string;
revision: number | null;
checked_at: string | null;
error: string | null;
}
export interface MaintenanceJob {
id: number;
server: string;
state: 'queued' | 'restarting' | 'done' | 'failed' | 'cancelled';
target_revision: number | null;
requested_at: string;
restarted_at: string | null;
finished_at: string | null;
detail: string | null;
}
export interface PlutoniumStatus {
release: PlutoniumRelease;
schedule: MaintenanceSchedule;
servers: PlutoniumVersion[];
jobs: MaintenanceJob[];
}
/** Accept explicit revision tokens only, never engine versions or build dates. */
export function parsePlutoniumRevision(text: string): number | null {
const match = text.replace(/\^[0-9]/g, '').match(/\br([1-9]\d{2,8})\b/i);
return match ? Number(match[1]) : null;
}
export function parseChangelogRevision(html: string): number {
const headings = [...html.matchAll(/<h2\b[^>]*>([\s\S]*?)<\/h2>/gi)];
const revisions = headings.map(h => parsePlutoniumRevision(h[1].replace(/<[^>]*>/g, ' ')))
.filter((r): r is number => r !== null);
if (!revisions.length) throw new Error('The Plutonium changelog did not contain a release heading.');
return Math.max(...revisions);
}
export function nextMaintenanceAt(time: string, now = new Date()): string {
if (!/^([01]\d|2[0-3]):[0-5]\d$/.test(time)) throw new Error('Choose a valid UTC time (HH:mm).');
const next = new Date(now);
const [hour, minute] = time.split(':').map(Number);
next.setUTCHours(hour, minute, 0, 0);
if (next <= now) next.setUTCDate(next.getUTCDate() + 1);
return next.toISOString();
}
+18
View File
@@ -362,6 +362,24 @@ CREATE TABLE IF NOT EXISTS settings (
key TEXT PRIMARY KEY,
value TEXT NOT NULL
);
CREATE TABLE IF NOT EXISTS plutonium_versions (
server TEXT PRIMARY KEY REFERENCES servers(name) ON DELETE CASCADE,
revision INTEGER,
checked_at TEXT NOT NULL,
error TEXT
);
CREATE TABLE IF NOT EXISTS maintenance_jobs (
id INTEGER PRIMARY KEY AUTOINCREMENT,
server TEXT NOT NULL,
state TEXT NOT NULL DEFAULT 'queued',
target_revision INTEGER,
requested_at TEXT NOT NULL,
restarted_at TEXT,
finished_at TEXT,
detail TEXT
);
CREATE UNIQUE INDEX IF NOT EXISTS idx_maintenance_pending ON maintenance_jobs(server)
WHERE state IN ('queued', 'restarting');
CREATE TABLE IF NOT EXISTS server_telemetry (
server TEXT PRIMARY KEY REFERENCES servers(name) ON DELETE CASCADE,
round INTEGER NOT NULL,
+173
View File
@@ -0,0 +1,173 @@
import { db } from '../db/index.js';
import { config, komodoConfigured, rconConfigured } from '../config.js';
import { getServer, listServers } from './servers.js';
import { rconGetPlayerCount, rconPlutoniumVersion } from '../integrations/rcon.js';
import { refreshRestartRollout, restartInRollout, rolloutHealth, rolloutWaitReason } from '../pipeline/restart-rollout.js';
import { nextMaintenanceAt, parseChangelogRevision } from '../../plutonium.js';
import type { MaintenanceJob, MaintenanceSchedule, PlutoniumRelease, PlutoniumStatus, PlutoniumVersion } from '../../plutonium.js';
const CHANGELOG = 'https://plutonium.pw/docs/changelog/';
const HOUR = 60 * 60 * 1000;
const nowIso = () => new Date().toISOString();
function readSetting<T>(key: string, fallback: T): T {
const row = db().prepare('SELECT value FROM settings WHERE key = ?').get(key) as { value: string } | undefined;
return row ? JSON.parse(row.value) : fallback;
}
function writeSetting(key: string, value: unknown): void {
db().prepare('INSERT INTO settings (key, value) VALUES (?, ?) ON CONFLICT(key) DO UPDATE SET value = excluded.value')
.run(key, JSON.stringify(value));
}
function release(): PlutoniumRelease {
return readSetting('plutonium_release', { revision: null, checkedAt: null, error: null });
}
function schedule(): MaintenanceSchedule {
return readSetting('plutonium_schedule', { enabled: false, time: '06:00', nextAt: null });
}
export function saveMaintenanceSchedule(enabled: boolean, time: string): void {
const nextAt = nextMaintenanceAt(time);
writeSetting('plutonium_schedule', { enabled, time, nextAt: enabled ? nextAt : null });
}
export function plutoniumStatus(): PlutoniumStatus {
const versions = db().prepare('SELECT * FROM plutonium_versions').all() as PlutoniumVersion[];
return {
release: release(), schedule: schedule(),
servers: listServers().map(s => versions.find(v => v.server === s.name)
?? { server: s.name, revision: null, checked_at: null, error: null }),
jobs: db().prepare(`SELECT * FROM maintenance_jobs WHERE state IN ('queued','restarting')
OR id IN (SELECT id FROM maintenance_jobs ORDER BY id DESC LIMIT 100) ORDER BY id DESC`).all() as MaintenanceJob[]
};
}
async function probeVersion(server: string): Promise<number | null> {
const recovering = db().prepare('SELECT 1 FROM restart_rollout WHERE server = ? AND healthy_at IS NULL').get(server);
const result = recovering ? { revision: null, error: 'Waiting for restart recovery.' }
: await rconPlutoniumVersion(server).catch(() => ({ revision: null, error: 'Version check failed.' }));
// Clear the old revision on failure: it is no longer evidence of what is running.
db().prepare(`INSERT INTO plutonium_versions (server, revision, checked_at, error) VALUES (?, ?, ?, ?)
ON CONFLICT(server) DO UPDATE SET revision = excluded.revision, checked_at = excluded.checked_at, error = excluded.error`)
.run(server, result.revision, nowIso(), result.error);
return result.revision;
}
let checking: Promise<void> | null = null;
export function checkPlutonium(): Promise<void> {
if (checking) return checking;
checking = check().finally(() => { checking = null; });
return checking;
}
async function check(): Promise<void> {
const previous = release();
try {
const response = await fetch(CHANGELOG, { signal: AbortSignal.timeout(config.httpTimeoutMs), redirect: 'error' });
if (!response.ok) throw new Error(`Changelog request failed (HTTP ${response.status}).`);
const html = await response.text();
if (html.length > 2_000_000) throw new Error('Changelog response was too large.');
writeSetting('plutonium_release', { revision: parseChangelogRevision(html), checkedAt: nowIso(), error: null });
} catch (err) {
writeSetting('plutonium_release', { ...previous, checkedAt: nowIso(), error: (err as Error).message });
}
const servers = listServers();
for (let i = 0; i < servers.length; i += 4) {
await Promise.all(servers.slice(i, i + 4).map(s => probeVersion(s.name)));
}
}
export function queueMaintenance(servers: string[]): number {
const names = new Set(listServers().map(s => s.name));
if (!servers.length || servers.some(s => !names.has(s))) throw new Error('Choose existing fleet servers.');
return db().transaction(() => {
const latest = release();
// An old cached release is still displayed, but never used as a fresh target.
const target = !latest.error && latest.checkedAt && Date.now() - Date.parse(latest.checkedAt) < HOUR
? latest.revision : null;
let added = 0;
for (const server of new Set(servers)) {
added += db().prepare(`INSERT OR IGNORE INTO maintenance_jobs (server, target_revision, requested_at)
VALUES (?, ?, ?)`).run(server, target, nowIso()).changes;
}
return added;
})();
}
export function cancelMaintenance(id: number): boolean {
return db().prepare(`UPDATE maintenance_jobs SET state = 'cancelled', finished_at = ?, detail = 'Cancelled by operator'
WHERE id = ? AND state = 'queued'`).run(nowIso(), id).changes > 0;
}
function conflictingWork(server: string): boolean {
return !!db().prepare(`SELECT 1 FROM server_deployments WHERE server = ? AND state IN ('queued','running','awaiting-health')
UNION ALL SELECT 1 FROM deployment_servers WHERE server = ? AND restart_state IN ('pending','failed')
UNION ALL SELECT 1 FROM beta_servers bs JOIN betas b ON b.id = bs.beta_id
WHERE bs.server = ? AND b.state IN ('running','ready-promote') LIMIT 1`).get(server, server, server);
}
function finish(job: MaintenanceJob, state: 'done' | 'failed' | 'cancelled', detail: string) {
db().prepare('UPDATE maintenance_jobs SET state = ?, detail = ?, finished_at = ? WHERE id = ?')
.run(state, detail, nowIso(), job.id);
}
function hold(job: MaintenanceJob, detail: string) {
db().prepare('UPDATE maintenance_jobs SET detail = ? WHERE id = ?').run(detail, job.id);
}
let ticking: Promise<void> | null = null;
export function maintenanceTick(): Promise<void> {
if (ticking) return ticking;
ticking = runTick().finally(() => { ticking = null; });
return ticking;
}
async function runTick(): Promise<void> {
const latest = release();
if (!latest.checkedAt || Date.now() - Date.parse(latest.checkedAt) >= HOUR) await checkPlutonium();
db().transaction(() => {
const plan = schedule();
if (plan.enabled && plan.nextAt && plan.nextAt <= nowIso()) {
const names = listServers().map(s => s.name);
if (names.length) queueMaintenance(names);
// Coalesce missed days after downtime into one run; no restart backlog.
writeSetting('plutonium_schedule', { ...plan, nextAt: nextMaintenanceAt(plan.time) });
}
})();
await refreshRestartRollout();
const jobs = db().prepare(`SELECT * FROM maintenance_jobs WHERE state IN ('queued','restarting') ORDER BY id`).all() as MaintenanceJob[];
for (const job of jobs) {
try { await processJob(job); }
catch { hold(job, 'Maintenance check failed; waiting for the next check.'); }
}
db().prepare(`DELETE FROM maintenance_jobs WHERE finished_at < ? AND state IN ('done','failed','cancelled')`)
.run(new Date(Date.now() - 30 * 24 * HOUR).toISOString());
}
async function processJob(job: MaintenanceJob): Promise<void> {
const server = getServer(job.server);
if (!server || server.destroyed_at) return finish(job, 'cancelled', 'Server removed.');
if (job.state === 'restarting') {
const slot = db().prepare('SELECT requested_at FROM restart_rollout WHERE server = ?').get(job.server) as { requested_at: string } | undefined;
if (!slot || !job.restarted_at || slot.requested_at < job.restarted_at) {
return finish(job, 'failed', 'Restart submission was interrupted. Inspect the server before queuing another restart.');
}
const health = rolloutHealth(job.server);
if (!health.ok) return hold(job, health.reason);
const revision = await probeVersion(job.server);
if (job.target_revision && (revision === null || revision < job.target_revision)) {
return finish(job, 'failed', `Server recovered, but r${job.target_revision} was not verified (${revision === null ? 'version unknown' : `running r${revision}`}). Check Plutainer auto-update settings.`);
}
return finish(job, 'done', `Server recovered; ${revision === null ? 'running revision unknown' : `running r${revision}`}.`);
}
if (!komodoConfigured() || !rconConfigured()) return hold(job, 'Configure Komodo and RCON before restarting.');
if (conflictingWork(job.server)) return hold(job, 'Waiting for deployment or beta work on this server.');
const reason = rolloutWaitReason(job.server);
if (reason) return hold(job, reason);
const players = await rconGetPlayerCount(job.server);
if (players === null || players > 0) return hold(job, players === null ? 'Waiting for a fresh RCON player count.' : `Waiting for ${players} player(s) to leave.`);
// Recheck after async I/O; cancellation and lifecycle operations may have arrived.
const current = getServer(job.server);
if (!current || current.destroyed_at) return finish(job, 'cancelled', 'Server removed.');
if (conflictingWork(job.server)) return hold(job, 'Waiting for deployment or beta work on this server.');
const claimed = db().prepare(`UPDATE maintenance_jobs SET state = 'restarting', restarted_at = ?, detail = 'Restart requested'
WHERE id = ? AND state = 'queued'`).run(nowIso(), job.id);
if (!claimed.changes) return;
db().prepare('UPDATE plutonium_versions SET revision = NULL, error = ? WHERE server = ?')
.run('Waiting for restart recovery.', job.server);
// Persist intent before remote I/O. An ambiguous timeout must never cause an automatic second restart.
const result = await restartInRollout(job.server);
if (result.kind === 'queued') {
db().prepare(`UPDATE maintenance_jobs SET state = 'queued', restarted_at = NULL, detail = ? WHERE id = ?`).run(result.message, job.id);
} else if (result.kind === 'failed') finish(job, 'failed', result.message);
}
+12
View File
@@ -3,6 +3,7 @@ import type { Socket } from 'node:dgram';
import { config, komodoConfigured, rconConfigured } from '../config.js';
import { getServer, setServerField } from '../fleet/servers.js';
import { komodoResolveGamePort } from './komodo.js';
import { parsePlutoniumRevision } from '../../plutonium.js';
// Plutonium/CoD RCON is UDP with a 4-byte 0xff prefix. Requests look like
// \xff\xff\xff\xffrcon <password> <command>
@@ -451,3 +452,14 @@ function positiveCapacity(value: string | undefined): number | null {
const count = Number(value);
return Number.isInteger(count) && count > 0 ? count : null;
}
/** Query the running process, not the updater's on-disk files. */
export async function rconPlutoniumVersion(server: string): Promise<{ revision: number | null; error: string | null }> {
const { endpoint } = await endpointFor(server);
if (!endpoint) return { revision: null, error: 'No RCON endpoint available.' };
const response = await sendRcon(endpoint.host, endpoint.port, 'version');
if (!response.ok) return { revision: null, error: response.detail };
const body = stripPrefix(response.buf);
const revision = body?.trimStart().startsWith('print') ? parsePlutoniumRevision(body) : null;
return { revision, error: revision === null ? 'RCON version reply did not identify a Plutonium revision.' : null };
}
+3
View File
@@ -13,6 +13,7 @@ import { isProvisioning } from '../fleet/provisioning.js';
import { STATUS_REFRESH_MS } from '../../server-freshness.js';
import { recordMetrics } from '../fleet/metrics.js';
import { selectedZombieMap } from '../../map-aliases.js';
import { maintenanceTick } from '../fleet/plutonium.js';
const TICK_MS = 15_000;
// Cap concurrent UDP RCON calls so a 15-server fleet doesn't blast the
@@ -46,6 +47,8 @@ async function runTick(): Promise<void> {
} catch (err) {
console.error('[worker] beta-runner error:', err);
}
try { await maintenanceTick(); }
catch (err) { console.error('[worker] maintenance error:', err); }
}
/**
@@ -0,0 +1,108 @@
<script lang="ts">
import { onMount } from 'svelte';
import type { PlutoniumStatus } from '$lib/plutonium.js';
import { timeFormat } from '$ui/stores/time-zone.js';
import Button from './primitives/Button.svelte';
let data = $state<PlutoniumStatus | null>(null);
let busy = $state(false);
let error = $state('');
let notice = $state('');
let enabled = $state(false);
let time = $state('06:00');
let initialized = false;
async function refresh(action?: Record<string, unknown>) {
if (busy) return;
busy = true;
try {
const response = await fetch('/api/plutonium', action ? {
method: 'POST', headers: { 'Content-Type': 'application/json' }, body: JSON.stringify(action)
} : undefined);
const result = await response.json();
if (!response.ok) throw new Error(result.error ?? 'Could not load Plutonium status.');
data = result;
error = '';
if (!initialized) {
enabled = result.schedule.enabled; time = result.schedule.time; initialized = true;
}
if (action) notice = action.action === 'schedule' ? 'Schedule saved.'
: action.action === 'restart' ? 'Restart queued. Occupied servers will wait until empty.'
: action.action === 'cancel' ? 'Queued restart cancelled.' : 'Version checks finished.';
} catch (cause) { error = (cause as Error).message; }
finally { busy = false; }
}
function pending(server: string) {
return data?.jobs.some(j => j.server === server && ['queued', 'restarting'].includes(j.state));
}
onMount(() => {
void refresh();
const timer = setInterval(() => { if (!document.hidden) void refresh(); }, 5000);
return () => clearInterval(timer);
});
</script>
<section id="plutonium" aria-labelledby="plutonium-heading">
<div class="heading"><h2 id="plutonium-heading">Plutonium updates & restarts</h2>
<Button disabled={busy} onclick={() => refresh({ action: 'check' })}>Check versions</Button>
</div>
<p>Plutainer applies updates on startup when auto-update is enabled. Restarts wait for a fresh player count of zero and share the fleet’s rolling restart limit.</p>
{#if error}<p class="ui-error" role="alert">{error}</p>{/if}
{#if notice}<p role="status">{notice}</p>{/if}
{#if data}
<p>Latest published revision: <strong>{data.release.revision ? `r${data.release.revision}` : 'Unknown'}</strong>
· <a href="https://plutonium.pw/docs/changelog/" target="_blank" rel="noreferrer">Changelog</a>
{#if data.release.checkedAt} · Checked {$timeFormat.dateTime(data.release.checkedAt)}{/if}
</p>
{#if data.release.error}<p role="alert">Update check failed: {data.release.error} Any displayed release is cached.</p>{/if}
<form onsubmit={(event) => { event.preventDefault(); void refresh({ action: 'schedule', enabled, time }); }}>
<label class="toggle"><input type="checkbox" bind:checked={enabled} /> Daily fleet restart</label>
<label>Queue time (UTC)<input type="time" bind:value={time} required /></label>
<Button type="submit" disabled={busy}>Save schedule</Button>
</form>
<p>This restarts all fleet servers daily, even without a new release. Busy servers may restart later. Disabling the schedule stops future runs; cancel existing queued restarts below.</p>
{#if data.schedule.nextAt}<p>Next run: {$timeFormat.dateTime(data.schedule.nextAt)}</p>{/if}
<div class="table-wrap"><table>
<thead><tr><th>Server</th><th>Running revision</th><th>Update status</th><th>Action</th></tr></thead>
<tbody>{#each data.servers as server (server.server)}
<tr><td><a href={`/servers/${encodeURIComponent(server.server)}`}>{server.server}</a></td>
<td>{server.revision ? `r${server.revision}` : 'Unknown'}
{#if server.checked_at}<small>Checked {$timeFormat.dateTime(server.checked_at)}</small>{/if}
{#if server.error}<small>{server.error}</small>{/if}
</td>
<td>{server.revision === null || data.release.revision === null || data.release.error ? 'Unverified'
: server.revision < data.release.revision ? 'Update available' : server.revision > data.release.revision ? 'Ahead of changelog' : 'Up to date'}</td>
<td><Button size="sm" disabled={busy || pending(server.server)} onclick={() => refresh({ action: 'restart', servers: [server.server] })}>
{pending(server.server) ? 'Restart pending' : 'Queue restart'}
</Button></td></tr>
{:else}<tr><td colspan="4">No fleet servers yet.</td></tr>{/each}</tbody>
</table></div>
{#if data.servers.length}<Button disabled={busy || data.servers.every(s => pending(s.server))} onclick={() => refresh({ action: 'restart', servers: data!.servers.map(s => s.server) })}>Queue fleet restart</Button>{/if}
{#if data.jobs.length}
<h3>Restart activity</h3>
<ul>{#each data.jobs as job (job.id)}<li>
<strong>{job.server}</strong> · {job.state} · {$timeFormat.dateTime(job.requested_at)}
{#if job.target_revision} · Target r{job.target_revision}{/if}
{#if job.detail}<small>{job.detail}</small>{/if}
{#if job.state === 'queued'}<Button size="sm" disabled={busy} onclick={() => refresh({ action: 'cancel', id: job.id })}>Cancel</Button>{/if}
</li>{/each}</ul>
{/if}
{:else}<p role="status">{busy ? 'Loading maintenance status…' : 'Maintenance status unavailable.'}</p>{/if}
</section>
<style>
section { padding: 20px; border: 1px solid var(--border-strong); border-radius: var(--radius-md); background: var(--surface); scroll-margin-top: 90px; }
.heading, form { display: flex; flex-wrap: wrap; gap: 16px; align-items: center; justify-content: space-between; }
h2 { font-size: var(--text-md); margin: 0; }
h3 { font-size: var(--text-sm); margin-top: 24px; }
p, td, th, label, li { font-size: var(--text-sm); line-height: 1.6; }
p, small { color: var(--text-secondary); }
label { display: grid; gap: 6px; }
.toggle { display: flex; align-items: center; }
input[type='time'] { padding: 8px; color: var(--text-primary); background: var(--surface); border: 1px solid var(--border-strong); border-radius: var(--radius-sm); }
.table-wrap { overflow-x: auto; margin: 16px 0; }
table { width: 100%; border-collapse: collapse; text-align: left; }
th, td { padding: 10px; border-bottom: 1px solid var(--border-strong); }
small { display: block; }
ul { padding-left: 20px; }
li { margin-bottom: 12px; }
</style>
+27
View File
@@ -0,0 +1,27 @@
import { json } from '@sveltejs/kit';
import { cancelMaintenance, checkPlutonium, plutoniumStatus, queueMaintenance, saveMaintenanceSchedule } from '$lib/server/fleet/plutonium.js';
export function GET() { return json(plutoniumStatus()); }
export async function POST({ request }: { request: Request }) {
let body;
try { body = await request.json(); } catch { return json({ error: 'Invalid JSON.' }, { status: 400 }); }
if (!body || typeof body !== 'object') return json({ error: 'Invalid request.' }, { status: 400 });
if (body.action === 'check') {
await checkPlutonium();
} else if (body.action === 'schedule') {
if (typeof body.enabled !== 'boolean' || typeof body.time !== 'string' || !/^([01]\d|2[0-3]):[0-5]\d$/.test(body.time)) {
return json({ error: 'Provide enabled and a valid UTC time (HH:mm).' }, { status: 400 });
}
saveMaintenanceSchedule(body.enabled, body.time);
} else if (body.action === 'restart') {
if (!Array.isArray(body.servers) || !body.servers.length || body.servers.some((s: unknown) => typeof s !== 'string')) {
return json({ error: 'Choose servers to restart.' }, { status: 400 });
}
try { queueMaintenance(body.servers); }
catch (err) { return json({ error: (err as Error).message }, { status: 400 }); }
} else if (body.action === 'cancel') {
if (!Number.isSafeInteger(body.id) || !cancelMaintenance(body.id)) return json({ error: 'Only queued restarts can be cancelled.' }, { status: 409 });
} else return json({ error: 'Unknown action.' }, { status: 400 });
return json(plutoniumStatus());
}
+2
View File
@@ -6,6 +6,7 @@
import SectionHeader from '$ui/components/primitives/SectionHeader.svelte';
import Button from '$ui/components/primitives/Button.svelte';
import Badge from '$ui/components/primitives/Badge.svelte';
import PlutoniumMaintenance from '$ui/components/PlutoniumMaintenance.svelte';
let data = $state<SystemStatus | null>(null);
let loading = $state(false);
let error = $state('');
@@ -38,6 +39,7 @@
{#if data}<p>Example: {$timeFormat.dateTime(data.checkedAt)}</p>{/if}
{#if $timeZonePersistenceError}<p role="status">Browser storage is unavailable. Your timezone applies for this visit but cannot be saved.</p>{/if}
</section>
<PlutoniumMaintenance />
<div class="heading">
<div><h2>System status</h2><p>Check your connections and the prerequisites for a playable server.</p></div>
<Button loading={loading} onclick={() => refresh(true)}>Check connections</Button>