sgl-project/sglang · error · ValueError

--asr-max-buffer-seconds must be positive (got {cfg.asr_max_

Error message

--asr-max-buffer-seconds must be positive (got {cfg.asr_max_buffer_seconds}).

What it means

Raised by _handle_asr_validation when --asr-max-buffer-seconds is <= 0. This buffers streaming ASR (transcription) audio, so a non-positive buffer duration is meaningless and rejected before server start.

Source

Thrown at python/sglang/srt/server_args.py:9554

            self._declare(
                "_handle_dllm_inference",
                disaggregation_mode="null",
            )

        if cfg.enable_mixed_chunk:
            logger.warning(
                "Mixed chunked prefill is disabled because of using diffusion LLM inference."
            )
            self._declare(
                "_handle_dllm_inference",
                enable_mixed_chunk=False,
            )

    def _handle_asr_validation(self):
        """Validate transcription/ASR-specific server args."""
        cfg = resolving_view(self)
        if cfg.asr_max_buffer_seconds <= 0:
            raise ValueError(
                f"--asr-max-buffer-seconds must be positive "
                f"(got {cfg.asr_max_buffer_seconds})."
            )
        if cfg.asr_max_concurrent_sessions <= 0:
            raise ValueError(
                f"--asr-max-concurrent-sessions must be positive "
                f"(got {cfg.asr_max_concurrent_sessions})."
            )

    def _validate_prefill_decode_interval(self):
        cfg = resolving_view(self)
        if cfg.prefill_decode_interval < 0:
            raise ValueError("--prefill-decode-interval must be non-negative.")

    def _handle_other_validations(self):
        cfg = resolving_view(self)
        if cfg.default_chat_template_kwargs is not None and not isinstance(
            cfg.default_chat_template_kwargs, dict

View on GitHub (pinned to 0132848349)

Solutions

1. Set a positive value, e.g. --asr-max-buffer-seconds 5.0
2. If you meant minimal latency, use a small positive buffer like 0.5

Example fix

# before
--asr-max-buffer-seconds 0
# after
--asr-max-buffer-seconds 5.0

if args.asr_max_buffer_seconds is not None:
    assert args.asr_max_buffer_seconds > 0, 'asr_max_buffer_seconds must be > 0'

When it happens

Trigger: Launching with --asr-max-buffer-seconds 0 or a negative value while ASR/transcription features are configured.

Common situations: Trying to disable buffering by setting 0; typos or unit confusion (passing milliseconds where seconds expected).

Related errors

• --asr-max-concurrent-sessions must be positive (got {cfg.asr
• Decode context parallel size (--dcp-size / --decode-context-
• --mamba-max-states-per-path must be -1 (unlimited) or a posi
• --swa-full-tokens-ratio should be in range (0, 1.0].
• --prefill-decode-interval must be non-negative.

AI-assisted analysis of sgl-project/sglang@0132848349 (2026-08-28). Data as JSON: /api/errors/47eb2f6735cf47da. Report an issue: GitHub.