diff --git a/src/llama_stack/cli/stack/run.py b/src/llama_stack/cli/stack/run.py index 2882500ce..044ce49c9 100644 --- a/src/llama_stack/cli/stack/run.py +++ b/src/llama_stack/cli/stack/run.py @@ -127,7 +127,7 @@ class StackRun(Subcommand): config = StackRunConfig(**cast_image_name_to_string(replace_env_vars(config_contents))) port = args.port or config.server.port - host = config.server.host or ["::", "0.0.0.0"] + host = config.server.host or "0.0.0.0" # Set the config file in environment so create_app can find it os.environ["LLAMA_STACK_CONFIG"] = str(config_file) @@ -139,6 +139,7 @@ class StackRun(Subcommand): "lifespan": "on", "log_level": logger.getEffectiveLevel(), "log_config": logger_config, + "workers": config.server.workers, } keyfile = config.server.tls_keyfile diff --git a/src/llama_stack/core/datatypes.py b/src/llama_stack/core/datatypes.py index 95907adcf..2182ea4e5 100644 --- a/src/llama_stack/core/datatypes.py +++ b/src/llama_stack/core/datatypes.py @@ -473,6 +473,10 @@ class ServerConfig(BaseModel): "- true: Enable localhost CORS for development\n" "- {allow_origins: [...], allow_methods: [...], ...}: Full configuration", ) + workers: int = Field( + default=1, + description="Number of workers to use for the server", + ) class StackRunConfig(BaseModel):