The best way to upgrade vLLM is actually NOT to upgrade it, but to wipe it’s venv and install from scratch, in my opinion and based on my experimentation, just run:
deactivate
rm -rf .venv
uv venv --python 3.12 --seed --managed-python
source .venv/bin/activate
uv pip install vllm --torch-backend=auto
That’s it, Enjoy!
Sources: My experiments + Setup guidelines from https://docs.vllm.ai/en/latest/getting_started/installation/gpu/#using-vllm-metal
A quick command to show you mounts and volume mappings in your docker container:
docker inspect CONTAINER_NAME --format '{{ range .Mounts }}{{ .Source }} -> {{ .Destination }}{{ "\n" }}{{ end }}'
Sample response:
/dir1/file1.txt -> /app/data/file1.txt/dir1/file2.conf -> /app/data/file2.conf/dir1/dir2 -> /app/data/folder1
That’s it, Enjoy!
To add an MCP server to OpenCode, just follow these steps:
opencode.json config file for editing, e.g. nano .config/opencode/opencode.json on macOS.
"mcp": { "my-remote-mcp": { "type": "remote", "url": "https://my-mcp-server.com", "enabled": true, "headers": { "Authorization": "Bearer MY_API_KEY" } } }
If opencode.json already has data in it, make sure to paste the above after schema and to add a , after the last closing braces, or opencode will error out.
Sources:
Just a couple of environment variables and arguments, simply do the following:
export VLLM_LOGGING_LEVEL=DEBUG
export VLLM_DEBUG_LOG_API_SERVER_RESPONSE=TRUE
vllm serve \
--enable-log-requests \
--enable-log-outputs 2>&1 | tee vllm_server.log
That’s it, Enjoy!
Source:
Simply run curl -fsSL https://lmstudio.ai/install.sh | bash
After you install, don’t forget to source your .bashrc file via source .bashrc so that you can run the lms command.
To bring it up, run lms daemon up
After it’s up, you can download a model using lms get, like this:
lms get MODEL_LINK@MODEL_QUANTIZATION
e.g.
lms get https://lmstudio.ai/models/nvidia/nemotron-3-nano-omni@q8_0
To load a model into memory:
lms load "nemotron-3-nano-omni" --context-length 128000
To start the server:
lms server start --port 1234
That’s it, Enjoy!
Sources:
Just Some Techie Notes! 