EngineX/xc-llm-ascend
3
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

[Doc][KV Pool]Update Memcache local service config example: increase default world size to 256 and update description (#7025)

Browse Source 
### What this PR does / why we need it?
Update Memcache local service config example: increase default world
size to 256 and update the description for better clarity.

### Does this PR introduce _any_ user-facing change?

### How was this patch tested?

- vLLM version: v0.16.0
- vLLM main:
4034c3d32e

Signed-off-by: Pz1116 <zpbzpb123123@gmail.com>
This commit is contained in:
pz1116
2026-03-06 10:23:55 +08:00
committed by GitHub
parent a838a89630
commit a7820d20f4
1 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
docs/source/user_guide/feature_guide/kv_pool.md
View File

@@ -484,8 +484,8 @@ ock.mmc.tls.key.pass.path = /opt/ock/security/certs/client.passphrase
ock.mmc.tls.package.path = /opt/ock/security/libs/
ock.mmc.tls.decrypter.path =
# Total count of local service
ock.mmc.local_service.world_size = 32
# Total count of local service, including services that will be add in the future
ock.mmc.local_service.world_size = 256
# config store url, it will automatically modified to PodIP at Pod startup in HA scenario
# keep consistent with the same name configuration in mmc-meta.conf
ock.mmc.local_service.config_store_url = tcp://xx.xx.xx.xx:6000
@@ -533,7 +533,7 @@ ock.mmc.client.write_thread_pool.size = 2
* ock.mmc.meta_service_url：Configure the IP address and port number of the master node. The IP address and port number of the P node and D node can be the same.
* ock.mmc.local_service.config_store_url：Configure the IP address and port number of the master node. The IP address and port number of the P node and D node can be the same.
* ock.mmc.local_service.world_size：Total number of cards for starting services.
* ock.mmc.local_service.world_size：Total count of local service, including services that will be add in the future.
* ock.mmc.local_service.protocol：host_rdma (default), device_rdma (supported for A2 and A3 when device ROCE available, recommended for A2), device_sdma (supported for A3 when HCCS available, recommended for A3)
* ock.mmc.local_service.dram.size：Sets the size of the memory occupied by the master. The configured value is the size of the memory occupied by each card.
* To disable TLS authentication modification, set the following parameters to false:：ock.mmc.meta.ha.enable、ock.mmc.config_store.tls.enable