hsml.deployment_version #
DeploymentVersion #
One retained configuration version of a deployment.
A deployment keeps every version it ever had. Exactly one of them is active, and rolling back reactivates an earlier one without copying it. Only the predictor, transformer and model artifact settings belong to a version. The API protocol, request batching, inference logging, scheduling configuration and Knative mode live on the deployment itself and are the same for every version.
config_file property #
config_file: str | None
Model server configuration file of this version.
This is a base name inside the version directory, which stores it with a configfile- prefix, or an absolute HopsFS path for a file read from outside the version.
feature_logging property #
feature_logging: DeploymentLoggingConfig | None
Feature logging overrides of the predictor.
git_auto_redeploy property #
git_auto_redeploy: bool | None
Whether the deployment redeploys itself when the tracked branch moves.
git_current_commit property #
git_current_commit: str | None
Commit the deployment was last rolled to by git-sync.
model_name property #
model_name: str | None
Name of the model this version serves, or None for a model-less deployment.
predictor property #
predictor: str | None
Predictor script of this version.
This is a base name inside the version directory, which stores it with a predictor- prefix, or an absolute HopsFS path for a script read from outside the version.
predictor_env_vars property #
User environment variables of the predictor, as KEY=VALUE strings.
predictor_environment property #
predictor_environment: str | None
Name of the Python environment the predictor runs in.
predictor_environment_id property #
predictor_environment_id: int | None
Id of the Python environment the predictor runs in.
predictor_resources property #
predictor_resources: PredictorResources | None
Resources allocated to the predictor.
predictor_scaling_config property #
predictor_scaling_config: PredictorScalingConfig | None
Scaling configuration of the predictor.
tracing property #
tracing: DeploymentTracingConfig | None
Tracing configuration of an agent deployment.
transformer property #
transformer: str | None
Transformer script of this version, or None without a transformer.
This is a base name inside the version directory, which stores it with a transformer- prefix, or an absolute HopsFS path for a script read from outside the version.
transformer_env_vars property #
User environment variables of the transformer, as KEY=VALUE strings.
transformer_environment property #
transformer_environment: str | None
Name of the Python environment the transformer runs in.
transformer_environment_id property #
transformer_environment_id: int | None
Id of the Python environment the transformer runs in.
transformer_resources property #
transformer_resources: TransformerResources | None
Resources allocated to the transformer.
transformer_scaling_config property #
transformer_scaling_config: TransformerScalingConfig | None
Scaling configuration of the transformer.
updated property #
updated: str | None
When this version was last edited in place, or None if it never was.