[Bugfix][KV Offload] Check cgroup memory before SHM allocation (#54014)
Signed-off-by: Alex <jihui.huang@daocloud.io> Signed-off-by: AlexHuang <alex.tech.lab@outlook.com> Signed-off-by: Alex <alex.tech.lab@outlook.com> Co-authored-by: OpenAI Codex <codex@openai.com> Co-authored-by: Itay Etelis <92247226+Etelis@users.noreply.github.com>
This commit is contained in:
co-authored by
OpenAI Codex
Itay Etelis
parent
e0301afb2a
commit
0f8fa53acf
@@ -727,11 +727,15 @@ def _get_available_ram_bytes() -> int:
|
||||
|
||||
host_available = psutil.virtual_memory().available
|
||||
|
||||
from vllm.utils.cpu_resource_utils import get_cgroup_memory_limit
|
||||
from vllm.utils.cpu_resource_utils import (
|
||||
get_cgroup_memory_limit,
|
||||
get_cgroup_memory_usage,
|
||||
)
|
||||
|
||||
cgroup_limit, cgroup_usage = get_cgroup_memory_limit()
|
||||
cgroup_limit = get_cgroup_memory_limit()
|
||||
if cgroup_limit is None:
|
||||
return host_available
|
||||
cgroup_usage = get_cgroup_memory_usage()
|
||||
cgroup_available = (
|
||||
cgroup_limit if cgroup_usage is None else max(0, cgroup_limit - cgroup_usage)
|
||||
)
|
||||
|
||||
Reference in New Issue
Block a user