Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

fix: Handle value error when convert shmem to binary size #2972

Merged
merged 2 commits into from
Oct 25, 2024
Merged

fix: Handle value error when convert shmem to binary size #2972

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
c99e738
fix: Handle value error when convert shmem to binary size
fregataa Oct 24, 2024
336d7ee
add news fragment
fregataa Oct 24, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 2 additions & 0 deletions src/ai/backend/manager/defs.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -88,3 +88,5 @@ class LockID(enum.IntEnum):

DEFAULT_KEYPAIR_RESOURCE_POLICY_NAME: Final = "default"
DEFAULT_KEYPAIR_RATE_LIMIT: Final = 10000

DEFAULT_SHARED_MEMORY_SIZE: Final[str] = "64m"
20 changes: 15 additions & 5 deletions src/ai/backend/manager/registry.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -132,7 +132,7 @@
TooManySessionsMatched,
)
from .config import LocalConfig, SharedConfig
from .defs import DEFAULT_IMAGE_ARCH, DEFAULT_ROLE, INTRINSIC_SLOTS
from .defs import DEFAULT_IMAGE_ARCH, DEFAULT_ROLE, DEFAULT_SHARED_MEMORY_SIZE, INTRINSIC_SLOTS
from .exceptions import MultiAgentError, convert_to_status_data
from .models import (
AGENT_RESOURCE_OCCUPYING_KERNEL_STATUSES,
Expand Down Expand Up @@ -1128,10 +1128,20 @@ async def enqueue_session(
# We need to subtract the amount of shared memory from the memory limit of
# a container, since tmpfs including /dev/shm uses host-side kernel memory
# and cgroup's memory limit does not apply.
shmem = resource_opts.get("shmem", None)
if shmem is None:
shmem = labels.get("ai.backend.resource.preferred.shmem", "64m")
shmem = BinarySize.from_str(shmem)
raw_shmem: Optional[str] = resource_opts.get("shmem")
if raw_shmem is None:
raw_shmem = labels.get("ai.backend.resource.preferred.shmem")
if not raw_shmem:
# raw_shmem is None or empty string ("")
raw_shmem = DEFAULT_SHARED_MEMORY_SIZE
fregataa marked this conversation as resolved.
Show resolved Hide resolved
try:
shmem = BinarySize.from_str(raw_shmem)
except ValueError:
log.warning(
f"Failed to convert raw `shmem({raw_shmem})` "
f"to a decimal value. Fallback to default({DEFAULT_SHARED_MEMORY_SIZE})."
)
shmem = BinarySize.from_str(DEFAULT_SHARED_MEMORY_SIZE)
resource_opts["shmem"] = shmem
image_min_slots = copy.deepcopy(image_min_slots)
image_min_slots["mem"] += shmem
Expand Down
Loading