Implementation:Bentoml BentoML Deployment Terminate Delete

Overview

Deployment Terminate and Delete implements the Principle:Bentoml_BentoML_Deployment_Termination principle by providing two distinct functions: terminate() to stop a running deployment while preserving its record, and delete() to permanently remove a deployment.

API

• bentoml.deployment.terminate()
• bentoml.deployment.delete()

Source

src/bentoml/deployment.py:L301-348

Import

import bentoml

Signatures

terminate()

def terminate(
    name: str,
    cluster: str = None,
    wait: bool = False,
) -> Deployment

delete()

def delete(
    name: str,
    cluster: str = None,
) -> None

Key Parameters

Function	Parameter	Type	Default	Description
terminate	name	str	(required)	Name of the deployment to terminate
terminate	cluster	str	None	Cluster where the deployment runs
terminate	wait	bool	False	Block until termination completes
delete	name	str	(required)	Name of the deployment to delete
delete	cluster	str	None	Cluster where the deployment runs

Inputs and Outputs

terminate()

Inputs:

• Deployment name (required)
• Cluster name (optional, uses default if not specified)
• Wait flag (optional, controls blocking behavior)

Outputs:

• Deployment object in terminated state, with updated status reflecting the stopped service

delete()

Inputs:

• Deployment name (required)
• Cluster name (optional, uses default if not specified)

Outputs:

• None - The deployment is permanently removed

Usage Examples

Terminate a Deployment

import bentoml

# Terminate without waiting (returns immediately)
deployment = bentoml.deployment.terminate("my-llm-service")
print(f"Status: {deployment.status}")  # Status: terminating

# Terminate and wait for completion
deployment = bentoml.deployment.terminate(
    "my-llm-service",
    wait=True,
)
print(f"Status: {deployment.status}")  # Status: terminated

Delete a Deployment

import bentoml

# Permanently remove a deployment
bentoml.deployment.delete("my-old-service")
# Returns None; deployment is gone

Terminate Then Delete

import bentoml

# Two-phase cleanup: terminate first, then delete
deployment = bentoml.deployment.terminate("my-service", wait=True)
print(f"Terminated: {deployment.name}")

# After confirming termination, permanently remove
bentoml.deployment.delete("my-service")
print("Deployment deleted")

Specify Cluster

import bentoml

# Terminate a deployment in a specific cluster
deployment = bentoml.deployment.terminate(
    "my-service",
    cluster="gcp-us-central1",
    wait=True,
)

# Delete from a specific cluster
bentoml.deployment.delete(
    "my-service",
    cluster="gcp-us-central1",
)

CLI Usage

# Terminate a deployment
bentoml deployment terminate my-llm-service

# Terminate and wait
bentoml deployment terminate my-llm-service --wait

# Delete a deployment
bentoml deployment delete my-llm-service

# Specify cluster
bentoml deployment terminate my-service --cluster gcp-us-central1
bentoml deployment delete my-service --cluster gcp-us-central1

Metadata

Property	Value
Implementation	Deployment Terminate and Delete
API	bentoml.deployment.terminate(), .delete()
Source	src/bentoml/deployment.py:L301-348
Domain	ML_Serving, Cloud_Deployment, Operations
Workflow	BentoCloud_Deployment
Principle	Principle:Bentoml_BentoML_Deployment_Termination

Knowledge Sources

• BentoML
• BentoML Docs

Related Pages

• Principle:Bentoml_BentoML_Deployment_Termination
• Environment:Bentoml_BentoML_BentoCloud_Credentials