biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[1/N] Elastic EP Milestone 2 (#34861)

Browse Source 
Signed-off-by: Yongji Wu <wuyongji317@gmail.com>
Signed-off-by: Itay Alroy <ialroy@nvidia.com>
Signed-off-by: Tyler Michael Smith <tlrmchlsmth@gmail.com>
Signed-off-by: Ron Tourgeman <rtourgeman@nvidia.com>
Co-authored-by: Yongji Wu <wuyongji317@gmail.com>
Co-authored-by: Tyler Michael Smith <tlrmchlsmth@gmail.com>
Co-authored-by: Ron Tourgeman <rtourgeman@nvidia.com>
This commit is contained in:
Itay Alroy
2026-02-28 06:46:42 +02:00
committed by GitHub
parent 90805ff464
commit dea268336f
53 changed files with 3613 additions and 1016 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
vllm/envs.py
View File

@@ -243,6 +243,8 @@ if TYPE_CHECKING:
VLLM_LORA_DISABLE_PDL: bool = False
VLLM_ENABLE_CUDA_COMPATIBILITY: bool = False
VLLM_CUDA_COMPATIBILITY_PATH: str | None = None
VLLM_ELASTIC_EP_SCALE_UP_LAUNCH: bool = False
VLLM_ELASTIC_EP_DRAIN_REQUESTS: bool = False
def get_default_cache_root():
@@ -1617,6 +1619,16 @@ environment_variables: dict[str, Callable[[], Any]] = {
"VLLM_CUDA_COMPATIBILITY_PATH": lambda: os.environ.get(
"VLLM_CUDA_COMPATIBILITY_PATH", None
),
# Whether it is a scale up launch engine for elastic EP,
# Should only be set by EngineCoreClient.
"VLLM_ELASTIC_EP_SCALE_UP_LAUNCH": lambda: bool(
int(os.getenv("VLLM_ELASTIC_EP_SCALE_UP_LAUNCH", "0"))
),
# Whether to wait for all requests to drain before sending the
# scaling command in elastic EP.
"VLLM_ELASTIC_EP_DRAIN_REQUESTS": lambda: bool(
int(os.getenv("VLLM_ELASTIC_EP_DRAIN_REQUESTS", "0"))
),
}