diff --git a/docker/Dockerfile b/docker/Dockerfile index 89c93558b5f6..e6a413074540 100644 --- a/docker/Dockerfile +++ b/docker/Dockerfile @@ -11,7 +11,7 @@ ARG GRACE_BLACKWELL_DEEPEP_BRANCH=gb200_blog_part_2 ARG HOPPER_SBO_DEEPEP_COMMIT=9f2fc4b3182a51044ae7ecb6610f7c9c3258c4d6 ARG DEEPEP_COMMIT=9af0e0d0e74f3577af1979c9b9e1ac2cad0104ee ARG BUILD_AND_DOWNLOAD_PARALLEL=8 -ARG SGL_KERNEL_VERSION=0.4.1 +ARG SGL_KERNEL_VERSION=0.4.1.post1 ARG SGL_VERSION ARG USE_LATEST_SGLANG=0 ARG GDRCOPY_VERSION=2.5.1 diff --git a/python/pyproject.toml b/python/pyproject.toml index 17b2f4bae2cd..bfdcd2762957 100755 --- a/python/pyproject.toml +++ b/python/pyproject.toml @@ -59,7 +59,7 @@ dependencies = [ "sentencepiece", "setproctitle", "flash-attn-4>=4.0.0b9", - "sglang-kernel==0.4.1", + "sglang-kernel==0.4.1.post1", "soundfile==0.13.1", "tiktoken", "timm==1.0.16", diff --git a/python/sglang/srt/entrypoints/engine.py b/python/sglang/srt/entrypoints/engine.py index 06b57b63d167..05723f0e1492 100644 --- a/python/sglang/srt/entrypoints/engine.py +++ b/python/sglang/srt/entrypoints/engine.py @@ -1170,7 +1170,7 @@ def _set_envs_and_config(server_args: ServerArgs): if _is_cuda: assert_pkg_version( "sglang-kernel", - "0.4.1", + "0.4.1.post1", "Please reinstall the latest version with `pip install sglang-kernel --force-reinstall`", ) diff --git a/test/manual/4-gpu-models/test_qwen35_hicache.py b/test/registered/4-gpu-models/test_qwen35_hicache.py similarity index 95% rename from test/manual/4-gpu-models/test_qwen35_hicache.py rename to test/registered/4-gpu-models/test_qwen35_hicache.py index 05737b252397..66b6cd9f3d31 100644 --- a/test/manual/4-gpu-models/test_qwen35_hicache.py +++ b/test/registered/4-gpu-models/test_qwen35_hicache.py @@ -1,8 +1,3 @@ -""" -# TODO: Fails on cu13 venv migration. Ref: https://github.com/sgl-project/sglang/actions/runs/24616960626/job/71980705674?pr=23119 -# Should move back to registered test after it's fixed -""" - import shutil import tempfile import unittest diff --git a/test/manual/hicache/test_hicache_storage.py b/test/registered/hicache/test_hicache_storage.py similarity index 87% rename from test/manual/hicache/test_hicache_storage.py rename to test/registered/hicache/test_hicache_storage.py index 76278689e3cd..112178a7611d 100644 --- a/test/manual/hicache/test_hicache_storage.py +++ b/test/registered/hicache/test_hicache_storage.py @@ -3,11 +3,6 @@ register_cuda_ci(est_time=99, suite="stage-b-test-1-gpu-small") register_amd_ci(est_time=300, suite="stage-b-test-1-gpu-small-amd") -""" -# TODO: Segmentation fault occurs when upgraded to Cu13. Ref: https://github.com/sgl-project/sglang/actions/runs/24603159715/job/71945537414?pr=23119") -# Should move back to registered test after it's fixed -""" - import time import unittest diff --git a/test/manual/hicache/test_hicache_storage_3fs_backend.py b/test/registered/hicache/test_hicache_storage_3fs_backend.py similarity index 94% rename from test/manual/hicache/test_hicache_storage_3fs_backend.py rename to test/registered/hicache/test_hicache_storage_3fs_backend.py index 6baa0e15249c..8bf69623d44e 100644 --- a/test/manual/hicache/test_hicache_storage_3fs_backend.py +++ b/test/registered/hicache/test_hicache_storage_3fs_backend.py @@ -2,8 +2,6 @@ Benchmark tests for HiCache Storage with 3FS backend. Usage: python3 -m pytest test/registered/hicache/test_hicache_storage_3fs_backend.py -v -# TODO: Segmentation fault occurs when upgraded to Cu13. Ref: https://github.com/sgl-project/sglang/actions/runs/24603159715/job/71945537414?pr=23119") -# Should move back to registered test after it's fixed """ import json diff --git a/test/manual/hicache/test_hicache_storage_file_backend.py b/test/registered/hicache/test_hicache_storage_file_backend.py similarity index 98% rename from test/manual/hicache/test_hicache_storage_file_backend.py rename to test/registered/hicache/test_hicache_storage_file_backend.py index 66cf3f0d6d9e..99fd26b4036b 100644 --- a/test/manual/hicache/test_hicache_storage_file_backend.py +++ b/test/registered/hicache/test_hicache_storage_file_backend.py @@ -2,8 +2,6 @@ E2E tests for HiCache Storage functionality. Usage: python3 -m pytest test/registered/hicache/test_hicache_storage_file_backend.py -v -# TODO: Segmentation fault occurs when upgraded to Cu13. Ref: https://github.com/sgl-project/sglang/actions/runs/24603159715/job/71945537414?pr=23119") -# Should move back to registered test after it's fixed """ import json diff --git a/test/manual/hicache/test_hicache_storage_mooncake_backend.py b/test/registered/hicache/test_hicache_storage_mooncake_backend.py similarity index 98% rename from test/manual/hicache/test_hicache_storage_mooncake_backend.py rename to test/registered/hicache/test_hicache_storage_mooncake_backend.py index c13fbff9049c..1a9b75a398dd 100644 --- a/test/manual/hicache/test_hicache_storage_mooncake_backend.py +++ b/test/registered/hicache/test_hicache_storage_mooncake_backend.py @@ -4,9 +4,6 @@ python3.10 -m pytest test/registered/hicache/test_hicache_storage_mooncake_backend.py -v """ -# TODO: Segmentation fault occurs when upgraded to Cu13. Ref: https://github.com/sgl-project/sglang/actions/runs/24601791606/job/71942123195?pr=23119") -# Should move back to registered test after it's fixed - import os import subprocess import time @@ -15,6 +12,7 @@ import requests from test_hicache_storage_file_backend import HiCacheStorageBaseMixin +from sglang.test.ci.ci_register import register_cuda_ci from sglang.test.test_utils import ( DEFAULT_MLA_MODEL_NAME_FOR_TEST, CustomTestCase, @@ -22,6 +20,8 @@ is_in_ci, ) +register_cuda_ci(est_time=236, suite="stage-b-test-2-gpu-large") + class HiCacheStorageMooncakeBackendBaseMixin(HiCacheStorageBaseMixin): """Base mixin class with common setup and utilities""" diff --git a/test/manual/hicache/test_hicache_storage_runtime_attach_detach.py b/test/registered/hicache/test_hicache_storage_runtime_attach_detach.py similarity index 98% rename from test/manual/hicache/test_hicache_storage_runtime_attach_detach.py rename to test/registered/hicache/test_hicache_storage_runtime_attach_detach.py index 64c6a471a116..1517d6d4f46f 100644 --- a/test/manual/hicache/test_hicache_storage_runtime_attach_detach.py +++ b/test/registered/hicache/test_hicache_storage_runtime_attach_detach.py @@ -7,8 +7,6 @@ Usage: python3 -m pytest test/registered/hicache/test_hicache_storage_runtime_attach_detach.py -v -# TODO: Segmentation fault occurs when upgraded to Cu13. Ref: https://github.com/sgl-project/sglang/actions/runs/24603159715/job/71945537414?pr=23119") -# Should move back to registered test after it's fixed """ import json diff --git a/test/manual/hicache/test_hicache_variants.py b/test/registered/hicache/test_hicache_variants.py similarity index 94% rename from test/manual/hicache/test_hicache_variants.py rename to test/registered/hicache/test_hicache_variants.py index b48a63d6bb80..c769cf40d0d6 100644 --- a/test/manual/hicache/test_hicache_variants.py +++ b/test/registered/hicache/test_hicache_variants.py @@ -5,8 +5,6 @@ """ Consolidated HiCache variant tests. Tests HiCache with different configurations: standard, MLA, EAGLE, and page size variants. -# TODO: Segmentation fault occurs when upgraded to Cu13. Ref: https://github.com/sgl-project/sglang/actions/runs/24603159715/job/71945537414?pr=23119") -# Should move back to registered test after it's fixed """ import unittest