kv_transfer: Rename the shared storage connectors (#30201)

Signed-off-by: Or Ozeri <oro@il.ibm.com>
This commit is contained in:
Or Ozeri
2025-12-08 20:46:09 -08:00
committed by GitHub
parent 03b91f7262
commit 4c6fd25880
27 changed files with 129 additions and 129 deletions
@@ -32,7 +32,7 @@ class MMMeta:
@dataclass
class ECSharedStorageConnectorMetadata(ECConnectorMetadata):
class ECExampleConnectorMetadata(ECConnectorMetadata):
mm_datas: list[MMMeta]
def __init__(self):
@@ -42,7 +42,7 @@ class ECSharedStorageConnectorMetadata(ECConnectorMetadata):
self.mm_datas.append(mm_data)
class ECSharedStorageConnector(ECConnectorBase):
class ECExampleConnector(ECConnectorBase):
# NOTE: This is Simple debug implementation of the EC connector.
# It save / load the EC cache to / from the disk.
@@ -76,7 +76,7 @@ class ECSharedStorageConnector(ECConnectorBase):
# Get the metadata
metadata: ECConnectorMetadata = self._get_connector_metadata()
assert isinstance(metadata, ECSharedStorageConnectorMetadata)
assert isinstance(metadata, ECExampleConnectorMetadata)
assert encoder_cache is not None
if metadata is None:
logger.warning(
@@ -160,7 +160,7 @@ class ECSharedStorageConnector(ECConnectorBase):
Args:
scheduler_output (SchedulerOutput): the scheduler output object.
"""
meta = ECSharedStorageConnectorMetadata()
meta = ECExampleConnectorMetadata()
for mm_hash, num_encoder_token in self._mm_datas_need_loads.items():
meta.add_mm_data(MMMeta.make_meta(mm_hash, num_encoder_token))
self._mm_datas_need_loads.clear()
@@ -79,7 +79,7 @@ class ECConnectorFactory:
# only load the files corresponding to the current connector.
ECConnectorFactory.register_connector(
"ECSharedStorageConnector",
"vllm.distributed.ec_transfer.ec_connector.shared_storage_connector",
"ECSharedStorageConnector",
"ECExampleConnector",
"vllm.distributed.ec_transfer.ec_connector.example_connector",
"ECExampleConnector",
)
@@ -144,9 +144,9 @@ class KVConnectorFactory:
# only load the files corresponding to the current connector.
KVConnectorFactory.register_connector(
"SharedStorageConnector",
"vllm.distributed.kv_transfer.kv_connector.v1.shared_storage_connector",
"SharedStorageConnector",
"ExampleConnector",
"vllm.distributed.kv_transfer.kv_connector.v1.example_connector",
"ExampleConnector",
)
KVConnectorFactory.register_connector(
@@ -65,7 +65,7 @@ class ReqMeta:
@dataclass
class SharedStorageConnectorMetadata(KVConnectorMetadata):
class ExampleConnectorMetadata(KVConnectorMetadata):
requests: list[ReqMeta] = field(default_factory=list)
def add_request(
@@ -81,7 +81,7 @@ class SharedStorageConnectorMetadata(KVConnectorMetadata):
)
class SharedStorageConnector(KVConnectorBase_V1):
class ExampleConnector(KVConnectorBase_V1):
# NOTE: This is Simple debug implementation of the KV connector.
# It save / load the KV cache to / from the disk.
# It does extra work which will overwrite the existing prefix-cache in GPU
@@ -157,7 +157,7 @@ class SharedStorageConnector(KVConnectorBase_V1):
# Get the metadata
metadata: KVConnectorMetadata = self._get_connector_metadata()
assert isinstance(metadata, SharedStorageConnectorMetadata)
assert isinstance(metadata, ExampleConnectorMetadata)
if metadata is None:
logger.warning(
@@ -241,7 +241,7 @@ class SharedStorageConnector(KVConnectorBase_V1):
return layer.reshape(2, num_pages * page_size, -1)[:, slot_mapping, ...]
connector_metadata = self._get_connector_metadata()
assert isinstance(connector_metadata, SharedStorageConnectorMetadata)
assert isinstance(connector_metadata, ExampleConnectorMetadata)
for request in connector_metadata.requests:
if request.is_store:
filename = self._generate_filename_debug(
@@ -315,7 +315,7 @@ class SharedStorageConnector(KVConnectorBase_V1):
Args:
scheduler_output (SchedulerOutput): the scheduler output object.
"""
meta = SharedStorageConnectorMetadata()
meta = ExampleConnectorMetadata()
total_need_load = 0
for new_req in scheduler_output.scheduled_new_reqs: