sgl-project/sglang · error · RuntimeError

Network error after {consecutive_errors} consecutive failure

Error message

Network error after {consecutive_errors} consecutive failures: {str(e)}

What it means

RuntimeError raised in generate_video when non-ConnectionError requests exceptions (timeouts, protocol errors, etc.) accumulate to max_consecutive_errors during status polling. The loop tolerates individual transient errors but gives up after the budget is spent.

Source

Thrown at python/sglang/multimodal_gen/apps/ComfyUI_SGLDiffusion/core/server_api.py:348

                        error_msg = (
                            error.get("message", "Unknown error")
                            if error
                            else "Unknown error"
                        )
                        raise RuntimeError(f"Video generation failed: {error_msg}")
                except requests.exceptions.ConnectionError as e:
                    # Connection errors - likely server is down
                    consecutive_errors += 1
                    if consecutive_errors >= max_consecutive_errors:
                        raise RuntimeError(
                            f"Lost connection to server after {consecutive_errors} consecutive errors. "
                            f"Server may be unavailable: {str(e)}"
                        )
                except requests.exceptions.RequestException as e:
                    # Other network errors - continue polling but track errors
                    consecutive_errors += 1
                    if consecutive_errors >= max_consecutive_errors:
                        raise RuntimeError(
                            f"Network error after {consecutive_errors} consecutive failures: {str(e)}"
                        )

                time.sleep(poll_interval)

            raise TimeoutError(
                f"Video generation timed out after {max_wait_time} seconds"
            )
        except requests.exceptions.RequestException as e:
            raise RuntimeError(f"Failed to generate video: {str(e)}")

    def _build_image_common_params(
        self,
        prompt: str,
        size: str,
        n: int,
        response_format: str,
        negative_prompt: Optional[str] = None,

View on GitHub (pinned to 0132848349)

Solutions

Increase the per-request timeout and/or poll_interval passed to generate_video.
If a proxy sits in front, raise its read timeout for the status endpoint.
Check server health/logs to see why responses are failing.
Raise max_consecutive_errors to ride out transient bursts.

Example fix

// before
result = api.generate_video(prompt=p)  # Network error after N consecutive failures
// after
result = api.generate_video(prompt=p, poll_interval=10, max_consecutive_errors=30)

Defensive patterns

Strategy: retry

Validate before calling

api.generate_video(prompt=p, poll_interval=10, max_consecutive_errors=30)  # pre-tune budgets

Try / catch

for attempt in range(2):
    try:
        return api.generate_video(prompt=p, poll_interval=10)
    except RuntimeError as e:
        if 'Network error after' not in str(e) or attempt: raise
        time.sleep(10)

Prevention

Use a poll_interval of 5-10s, not sub-second.
Raise proxy read timeouts for status endpoints.
Size max_consecutive_errors to expected transient bursts.

When it happens

Trigger: Poll requests repeatedly timing out (per-request timeout too small for a busy server) or malformed responses for the whole consecutive-error budget while the job polls.

Common situations: Server so busy generating that poll responses are slow; reverse proxy with aggressive timeouts between client and server; poll_interval too aggressive hammering the server.

Related errors

AI-assisted analysis of sgl-project/sglang@0132848349 (2026-08-28). Data as JSON: /api/errors/49c5a9fcad0f861b. Report an issue: GitHub.