Cloudflare Error 524 means Cloudflare connected to your origin server, but the server didn’t respond with an HTTP response within 100 seconds.

Why this happens

Cloudflare acts as a reverse proxy between users and your origin server. It enforces a 100-second timeout on the connection to your origin (non-Enterprise plans). If your server takes longer than that to generate a response — whether due to a slow database query, heavy computation, or an overloaded process — Cloudflare drops the connection and shows a 524 error to the user.

What causes this error

1. Slow origin server — your app takes too long to process the request
2. Long-running process — file uploads, report generation, database migrations
3. Origin server overloaded — too many requests, not enough resources

Fix 1: Optimize the slow endpoint

Check which endpoint is timing out and optimize it:

• Add database indexes for slow queries
• Move heavy processing to background jobs
• Cache expensive computations

Fix 2: Increase Cloudflare timeout (Enterprise only)

Enterprise plans can increase the proxy read timeout up to 600 seconds. Free/Pro/Business plans are fixed at 100 seconds.

Fix 3: Use background processing

For long-running tasks, return a 202 Accepted immediately and process in the background:

app.post("/generate-report", (req, res) => {
  const jobId = startBackgroundJob(req.body);
  res.status(202).json({ jobId, status: "processing" });
});

app.get("/job/:id", (req, res) => {
  const status = getJobStatus(req.params.id);
  res.json(status);
});

Fix 4: Bypass Cloudflare for long requests

Use a subdomain that’s not proxied through Cloudflare (grey cloud in DNS settings) for endpoints that legitimately take over 100 seconds.

Alternative solutions

Use WebSockets or Server-Sent Events (SSE) for long-running operations — Cloudflare has a separate, longer timeout for WebSocket connections. You can also use Cloudflare Workers to stream partial responses back to the client, keeping the connection alive while your origin processes the request.

Prevention

• Set up monitoring and alerts for endpoint response times so you catch slow endpoints before they hit the 100-second limit.
• Design APIs so that any operation expected to take more than a few seconds uses an async job pattern (submit → poll → retrieve) instead of blocking the HTTP response.

Related: Nginx: 504 Gateway Timeout fix · Vercel Serverless Timeout fix · Cloudflare Workers vs Vercel Edge