refactor(disagg): register SGLANG_ENCODER_MM_LOAD_WORKERS in Envs (#36006)

This commit is contained in:
Shangming Cai
2026-08-23 12:47:38 +08:00
committed by GitHub
parent 97ae27893a
commit bd72190728
2 changed files with 2 additions and 2 deletions
@@ -10,7 +10,6 @@ import asyncio
import concurrent.futures import concurrent.futures
import functools import functools
import logging import logging
import os
from dataclasses import dataclass from dataclasses import dataclass
from typing import Callable, List, Optional, Tuple, Union from typing import Callable, List, Optional, Tuple, Union
@@ -149,7 +148,7 @@ class EncoderPreprocessor:
max_workers=envs.SGLANG_ENCODER_PREPROC_WORKERS.get() max_workers=envs.SGLANG_ENCODER_PREPROC_WORKERS.get()
) )
self.io_executor = concurrent.futures.ThreadPoolExecutor( self.io_executor = concurrent.futures.ThreadPoolExecutor(
max_workers=int(os.environ.get("SGLANG_ENCODER_MM_LOAD_WORKERS", 4)) max_workers=envs.SGLANG_ENCODER_MM_LOAD_WORKERS.get()
) )
# ------------------------------------------------------------------ # ------------------------------------------------------------------
+1
View File
@@ -1235,6 +1235,7 @@ class Envs:
SGLANG_ENCODER_IMAGE_PROCESSOR_USE_GPU = EnvBool(False) SGLANG_ENCODER_IMAGE_PROCESSOR_USE_GPU = EnvBool(False)
SGLANG_ENCODER_MAX_BATCH_SIZE = EnvInt(8) SGLANG_ENCODER_MAX_BATCH_SIZE = EnvInt(8)
SGLANG_ENCODER_PREPROC_WORKERS = EnvInt(8) SGLANG_ENCODER_PREPROC_WORKERS = EnvInt(8)
SGLANG_ENCODER_MM_LOAD_WORKERS = EnvInt(4)
# EncoderBootstrapServer health-check tuning. Interval == 0 disables it. # EncoderBootstrapServer health-check tuning. Interval == 0 disables it.
SGLANG_ENCODER_BOOTSTRAP_HEALTH_CHECK_INTERVAL = EnvFloat(10.0) SGLANG_ENCODER_BOOTSTRAP_HEALTH_CHECK_INTERVAL = EnvFloat(10.0)
SGLANG_ENCODER_BOOTSTRAP_HEALTH_CHECK_TIMEOUT = EnvFloat(2.0) SGLANG_ENCODER_BOOTSTRAP_HEALTH_CHECK_TIMEOUT = EnvFloat(2.0)