Rate limits

The Servor API enforces a rate limit per token so one integration can't starve the others. This page explains the exact limits, what a 429 response looks like, and how to handle it cleanly with the retry-after header.

The limits

Limits are applied per API token, not per team or per IP. Each token gets its own budget:

If you run several integrations, give each its own token — separate tokens get separate budgets, and you can revoke one without touching the rest. Manage your tokens in Servor.

When you hit the limit

Exceeding either window returns HTTP 429 Too Many Requests with a retry-after header telling you how many seconds to wait before retrying:

HTTP/1.1 429 Too Many Requests
retry-after: 2

{ "error": "Rate limit exceeded", "code": "rate_limited" }

Handling 429 in your code

The pattern is simple: on a 429, wait for retry-after seconds, then retry. Cap the number of retries so a persistent limit doesn't loop forever.

async function apiGet(path: string, token: string, tries = 5): Promise<Response> {
  const res = await fetch(`https://api.servor.app/v1${path}`, {
    headers: { Authorization: `Bearer ${token}` },
  });
 
  if (res.status === 429 && tries > 0) {
    const wait = Number(res.headers.get("retry-after") ?? "1");
    await new Promise((r) => setTimeout(r, wait * 1000));
    return apiGet(path, token, tries - 1);
  }
 
  return res;
}

# curl: retry automatically, honoring retry-after
curl --retry 5 --retry-delay 0 \
  https://api.servor.app/v1/monitors \
  -H "Authorization: Bearer sv_live_xxxxxxxxxxxx"

Staying under the limit

• Throttle your requests. Keep steady-state traffic under 3 requests/second per token rather than bursting.
• Poll on a sensible interval. For status mirrors and dashboards, poll every 30–60 seconds, not every second — you'll stay far under 5000/hour.
• Cache what rarely changes. Server and status-page metadata don't need to be re-fetched on every loop.
• Spread load across tokens. Independent workloads deserve independent tokens and independent budgets.
• Back off on 429. Treat it as normal flow control, not an error to log and ignore.

See also