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

[distributed][misc] error on same VLLM_HOST_IP setting (#7756)

Browse Source
This commit is contained in:
youkaichao 2024-08-21 16:25:34 -07:00 committed by GitHub
parent 8678a69ab5
commit 7eebe8ccaa
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 17 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
vllm/envs.py
View File

@ -137,7 +137,10 @@ environment_variables: Dict[str, Callable[[], Any]] = {
os.path.join(get_default_cache_root(), "vllm"), os.path.join(get_default_cache_root(), "vllm"),
)), )),
# used in distributed environment to determine the master address # used in distributed environment to determine the ip address
# of the current node, when the node has multiple network interfaces.
# If you are using multi-node inference, you should set this differently
# on each node.
'VLLM_HOST_IP': 'VLLM_HOST_IP':
lambda: os.getenv('VLLM_HOST_IP', "") or os.getenv("HOST_IP", ""), lambda: os.getenv('VLLM_HOST_IP', "") or os.getenv("HOST_IP", ""),

13
vllm/executor/ray_gpu_executor.py
View File

@ -218,6 +218,19 @@ class RayGPUExecutor(DistributedGPUExecutor):
for node_id, gpu_ids in node_gpus.items(): for node_id, gpu_ids in node_gpus.items():
node_gpus[node_id] = sorted(gpu_ids) node_gpus[node_id] = sorted(gpu_ids)
all_ips = set(worker_ips + [driver_ip])
n_ips = len(all_ips)
n_nodes = len(node_workers)
if n_nodes != n_ips:
raise RuntimeError(
f"Every node should have a unique IP address. Got {n_nodes}"
f" nodes with node ids {list(node_workers.keys())} and "
f"{n_ips} unique IP addresses {all_ips}. Please check your"
" network configuration. If you set `VLLM_HOST_IP` or "
"`HOST_IP` environment variable, make sure it is unique for"
" each node.")
VLLM_INSTANCE_ID = get_vllm_instance_id() VLLM_INSTANCE_ID = get_vllm_instance_id()
# Set environment variables for the driver and workers. # Set environment variables for the driver and workers.