Add server-side health checks for service online status
- update-health.js: pings every service URL from script.js (single source of truth) in parallel every minute via cron and writes health.json; any HTTP response below 500 counts as online, 5xx as error, timeouts/DNS failures as offline; TLS validity ignored - Card badges now show real state instead of a static Online label: grey Checking, green Online (with latency tooltip), amber Error (HTTP code), red Offline (reason); refreshed every 30 s - Entrypoint runs a first sweep at boot; nginx serves /health.json with no-cache; README documents the behavior
This commit is contained in:
@@ -0,0 +1,111 @@
|
||||
// ---------------------------------------------------------------------------
|
||||
// Health checks for all dashboard services.
|
||||
// Reads the service URLs from script.js (single source of truth), pings each
|
||||
// one in parallel and writes health.json. Scheduled every minute by cron.
|
||||
//
|
||||
// Status classification:
|
||||
// online - got any HTTP response below 500 (401/404 still means "up")
|
||||
// error - server responded with 5xx
|
||||
// offline - timeout / connection refused / DNS failure
|
||||
// TLS certificate validity is ignored (reachability check only).
|
||||
// ---------------------------------------------------------------------------
|
||||
|
||||
const http = require("http");
|
||||
const https = require("https");
|
||||
const fs = require("fs");
|
||||
const path = require("path");
|
||||
const { URL } = require("url");
|
||||
|
||||
const OUT_FILE = process.env.OUT_FILE || "/usr/share/nginx/html/health.json";
|
||||
const TIMEOUT = parseInt(process.env.HEALTH_TIMEOUT || "5000", 10);
|
||||
|
||||
// Reachability only — don't fail on self-signed certificates.
|
||||
const httpAgent = new http.Agent({ rejectUnauthorized: false });
|
||||
const httpsAgent = new https.Agent({ rejectUnauthorized: false });
|
||||
|
||||
function serviceUrls() {
|
||||
const candidates = [
|
||||
process.env.SERVICES_JS,
|
||||
"/usr/share/nginx/html/script.js",
|
||||
path.join(__dirname, "..", "script.js"),
|
||||
].filter(Boolean);
|
||||
for (const p of candidates) {
|
||||
try {
|
||||
const src = fs.readFileSync(p, "utf8");
|
||||
const urls = [...src.matchAll(/url: *"(https?:\/\/[^"]+)"/g)].map((m) => m[1]);
|
||||
if (urls.length) return [...new Set(urls)];
|
||||
} catch {
|
||||
// try next candidate
|
||||
}
|
||||
}
|
||||
return [];
|
||||
}
|
||||
|
||||
function checkUrl(url) {
|
||||
return new Promise((resolve) => {
|
||||
const started = Date.now();
|
||||
let settled = false;
|
||||
const done = (result) => {
|
||||
if (settled) return;
|
||||
settled = true;
|
||||
resolve(result);
|
||||
};
|
||||
|
||||
let req;
|
||||
try {
|
||||
const u = new URL(url);
|
||||
const client = u.protocol === "https:" ? https : http;
|
||||
req = client.request(
|
||||
u,
|
||||
{
|
||||
method: "GET",
|
||||
agent: u.protocol === "https:" ? httpsAgent : httpAgent,
|
||||
timeout: TIMEOUT,
|
||||
},
|
||||
(res) => {
|
||||
const code = res.statusCode;
|
||||
res.destroy(); // response headers are all we need
|
||||
done({
|
||||
status: code >= 500 ? "error" : "online",
|
||||
code,
|
||||
latencyMs: Date.now() - started,
|
||||
error: null,
|
||||
});
|
||||
}
|
||||
);
|
||||
} catch (e) {
|
||||
return done({ status: "offline", code: null, latencyMs: null, error: e.message });
|
||||
}
|
||||
|
||||
req.on("timeout", () => {
|
||||
req.destroy();
|
||||
done({ status: "offline", code: null, latencyMs: null, error: "Timeout" });
|
||||
});
|
||||
req.on("error", (e) => {
|
||||
done({ status: "offline", code: null, latencyMs: null, error: e.message });
|
||||
});
|
||||
req.end();
|
||||
});
|
||||
}
|
||||
|
||||
async function main() {
|
||||
const urls = serviceUrls();
|
||||
if (urls.length === 0) {
|
||||
console.error("No service URLs found — is script.js available?");
|
||||
process.exit(1);
|
||||
}
|
||||
|
||||
const results = await Promise.all(urls.map(async (u) => [u, await checkUrl(u)]));
|
||||
const services = Object.fromEntries(results);
|
||||
|
||||
const payload = { updatedAt: new Date().toISOString(), services };
|
||||
fs.writeFileSync(OUT_FILE, JSON.stringify(payload, null, 2));
|
||||
|
||||
const count = (s) => results.filter(([, r]) => r.status === s).length;
|
||||
console.log(`Health: ${urls.length} services — ${count("online")} online, ${count("error")} error, ${count("offline")} offline`);
|
||||
}
|
||||
|
||||
main().catch((err) => {
|
||||
console.error(err);
|
||||
process.exit(1);
|
||||
});
|
||||
Reference in New Issue
Block a user