Skip to content

bug: lmcache server points to wrong file in entrypoint #676

@yyzxw

Description

@yyzxw

Describe the bug

The current implementation of lmcache-server in deployment-cache-server.yaml when:

using the latest tag assumes that the executable file is located in /opt/venv/bin/lmcache_experimental_server
using the version tag assumes that the executable file is located in ~/lmcache_experimental_server

  - name: "lmcache-server"
  image: "{{ required "Required value 'cacheserverSpec.repository' must be defined !" .Values.cacheserverSpec.repository }}:{{ required "Required value 'cacheserverSpec.tag' must be defined !" .Values.cacheserverSpec.tag }}"
  command:
  {{- if eq .Values.cacheserverSpec.tag "latest-nightly" }}
  - "/opt/venv/bin/lmcache_server"
  {{- else if eq .Values.cacheserverSpec.tag "latest" }}
  - "/opt/venv/bin/lmcache_experimental_server"
  {{- else }}
  - "lmcache_experimental_server"
  {{- end }}
  - "0.0.0.0"
  - "{{ .Values.cacheserverSpec.containerPort }}"
  {{- if .Values.cacheserverSpec.resources }}

But in fact, starting from v0.3.0 of lmcache/vllm-openai the executable files are kept only in /opt/venv/bin and are named:

docker run --entrypoint /bin/sh  lmcache/vllm-openai:latest -c "ls /opt/venv/bin | grep lmcache"
lmcache_controller
lmcache_server
lmcache_v0_server

To Reproduce

docker run --entrypoint /bin/sh lmcache/vllm-openai:latest -c "ls /opt/venv/bin | grep lmcache"

Expected behavior

correctly identify the file

Additional context

No response

Metadata

Metadata

Assignees

Labels

bugSomething isn't working

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions