llama stack build and llama stack run are not symmetric for --image-name argument when venv is used

[booxter]

System Info

.

Information

• The official example scripts
• My own modified scripts

🐛 Describe the bug

When building and running with venv, I have to pass different --image-name. For build, it's always prepended with llamastack- prefix. For run, it's not. I think both commands should behave the same way, otherwise it's confusing.

$ llama stack build --template experimental-post-training --image-type venv --image-name train
$ llama stack run --image-type venv ~/.llama/distributions/experimental-post-training/experimental-post-training-run.yaml --image-name llamastack-train

Error logs

.

Expected behavior

Both commands should use the same --image-name format.

[booxter]

@cdoern FYI (Though not sure if this is specific to venv.)

[cdoern]

@booxter I will take a look at this 🙏

[JamesKunstle]

I just experienced the same thing, which was kinda confusing.

Steps:

1. python3.11 -m venv venv
2. source venv/bin/activate
3. pip3 install ./llama-stack
4. pip3 install ./llama-stack-client-python
5. llama stack build --template ollama --image-type venv --> creates llamastack-None environment

The venv that's created should be named exactly what the user puts in --image-name during the build phase, and something more verbose, like f"llamastack-{template-name}-{conda | venv | image}" by default if they put nothing.