stanfordnlp/CoreNLP · critical · TimeoutException

Never got readiness from annotator:

Error message

Never got readiness from annotator:

What it means

After the server is live, WebServiceAnnotator.ensureServer waits for the server's own readiness signal (server.ready / ready(true)). If readiness is never reported within CONNECT_TIMEOUT, it throws this TimeoutException.

Solutions

  1. Increase the connect/readiness timeout so slow initialization can complete
  2. Confirm the service's readiness/health endpoint actually flips to ready (curl the health path)
  3. Check service logs for initialization errors or deadlocks during warm-up
  4. Restart the service; if readiness is never signalled even when healthy, update the annotator configuration to point at the correct health URL

Example fix

// before
WebServiceAnnotator a = new WebServiceAnnotator(props); // default short timeout
// after
props.setProperty("webservice.connectTimeout", "300000");
WebServiceAnnotator a = new WebServiceAnnotator(props);
Defensive patterns

Strategy: retry

Validate before calling

// Poll the health endpoint until ready before using the annotator
long deadline = System.currentTimeMillis() + 300_000;
while (System.currentTimeMillis() < deadline) {
  try {
    int code = ((java.net.HttpURLConnection) new java.net.URL(healthUrl).openConnection()).getResponseCode();
    if (code == 200) break;
  } catch (java.io.IOException ignored) {}
  try { Thread.sleep(1000); } catch (InterruptedException ignored) {}
}

Try / catch

try {
  pipeline.annotate(doc);
} catch (java.util.concurrent.TimeoutException e) {
  if (e.getMessage().contains("Never got readiness")) {
    log.error("Server live but not ready in time: check health endpoint and init logs");
  }
}

Prevention

When it happens

Trigger: The server process is up and answering connections but its readiness check never returns true within the timeout — e.g. the service reports liveness before models finish loading, or the readiness endpoint is unreachable while the main endpoint is not.

Common situations: Slow model warm-up exceeding the timeout, a readiness endpoint path misconfigured, server started externally (no managed server object) so readiness must come from polling ready(false/true), or hung initialization in the service.

Understand the failure class

Background: Request timed out: what client-side request timeouts mean across libraries (Request timed out, TIMED_OUT, APITimeoutError) — this error's family across 39 libraries.

Related errors

AI-assisted analysis of stanfordnlp/CoreNLP@1b7edd19c4 (2026-09-10). Data as JSON: /api/errors/f1ea650ca27d9ae5. Report an issue: GitHub.

Appendix: source

Thrown at src/edu/stanford/nlp/pipeline/WebServiceAnnotator.java:278

      if (System.currentTimeMillis() > startTime + CONNECT_TIMEOUT) {
        throw new TimeoutException("Could not connect to annotator: " + this);
      }
      if (!live()) {
        try {
          Thread.sleep(1000);
        } catch (InterruptedException ignored) {}
      } else {
        everLive = true;
      }
    }
    log.info("Got liveness from server for " + this);

    // 3. Wait for the target server to become ready
    synchronized (this) {
      if (this.server.isPresent()) {
        while (!this.server.get().ready) {
          if (System.currentTimeMillis() > startTime + CONNECT_TIMEOUT) {
            throw new TimeoutException("Never got readiness from annotator: " + this);
          }
          if (!ready(true)) {
            try {
              Thread.sleep(1000);
            } catch (InterruptedException ignored) {
            }
          } else {
            this.server.get().ready = true;
          }
        }
      } else if (!ready(false)) { // The server is not ready
        throw new IOException("Server is not ready and can not start it!");
      }
    }
    log.info("Got readiness from server for " + this);
    serverWasActive = true;

    // 4. Server is ensured! We can continue

View on GitHub (pinned to 1b7edd19c4)