zenml deployment

The zenml deployment command group manages persistent pipeline deployments that run continuously or on schedules.

Overview

Deployments are different from regular pipeline runs:

Pipeline Runs: Execute once and finish
Deployments: Remain active and execute pipelines on triggers (schedules, events, webhooks)

Deployments enable:

Scheduled pipeline execution (daily, weekly, cron expressions)
Continuous inference pipelines
Webhook-triggered pipelines
Long-running model serving

Deployments require an orchestrator that supports them (e.g., Kubernetes, Kubeflow, Airflow).

Command Structure

zenml deployment <command> [OPTIONS] [ARGUMENTS]

Core Commands

List Deployments

View all registered deployments:

zenml deployment list [OPTIONS]

Default columns: id, name, status, url, pipeline, stack Filter options:

Option	Description
`--name`	Filter by deployment name
`--pipeline`	Filter by pipeline name or ID
`--stack`	Filter by stack name or ID
`--status`	Filter by status (active, inactive, failed)
`--user`	Filter by user
`--created`	Filter by creation date
`--columns`	Customize displayed columns
`--output`, `-o`	Output format (table, json, yaml, csv, tsv)
`--sort_by`	Sort results

Examples:

# List all deployments
zenml deployment list

# Filter by pipeline
zenml deployment list --pipeline training_pipeline

# Filter by status
zenml deployment list --status active

# Custom columns
zenml deployment list --columns id,name,status,created

# Export to JSON
zenml deployment list --output json

# Sort by creation date
zenml deployment list --sort_by "desc:created"

# Multiple filters
zenml deployment list --pipeline inference_pipeline --status active --stack production

Example output:

┏━━━━━━━━━━┯━━━━━━━━━━━━━━━━━━━━━┯━━━━━━━━━┯━━━━━━━━━━━━━━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━━━━━━┯━━━━━━━━━━━━┓
┃ ID       │ Name                │ Status  │ URL                       │ Pipeline           │ Stack      ┃
┠──────────┼─────────────────────┼─────────┼───────────────────────────┼────────────────────┼────────────┨
┃ 1a2b3c4d │ daily-training      │ active  │ https://dashboard/d/1a... │ training_pipeline  │ production ┃
┃ 5e6f7g8h │ inference-service   │ active  │ https://dashboard/d/5e... │ inference_pipeline │ production ┃
┃ 9i0j1k2l │ weekly-report       │ active  │ https://dashboard/d/9i... │ report_pipeline    │ staging    ┃
┗━━━━━━━━━━┷━━━━━━━━━━━━━━━━━━━━━┷━━━━━━━━━┷━━━━━━━━━━━━━━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━━━━━━┷━━━━━━━━━━━━┛

Describe a Deployment

View detailed information about a specific deployment:

zenml deployment describe <DEPLOYMENT_NAME_OR_ID> [OPTIONS]

Options:

Option	Description
`--show-secret`, `-s`	Display secret configuration values
`--show-metadata`, `-m`	Show deployment metadata
`--show-schema`, `-sc`	Show the deployment schema
`--no-truncate`, `-nt`	Don’t truncate long values

Examples:

# Basic description
zenml deployment describe daily-training

# Show all details
zenml deployment describe daily-training --show-metadata --show-schema

# Show secrets
zenml deployment describe daily-training --show-secret

# No truncation for long values
zenml deployment describe daily-training --show-metadata --no-truncate

Example output:

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
┃         Deployment: daily-training                ┃
┠───────────────────────────────────────────────────┨
┃ ID              │ 1a2b3c4d-e5f6-7890-abcd-...    ┃
┃ Name            │ daily-training                  ┃
┃ Status          │ active                          ┃
┃ Pipeline        │ training_pipeline               ┃
┃ Stack           │ production                      ┃
┃ Created         │ 2024-01-15 10:30:00            ┃
┃ Updated         │ 2024-03-09 08:00:00            ┃
┃ Schedule        │ 0 2 * * * (daily at 2 AM UTC)  ┃
┃ Dashboard URL   │ https://dashboard/d/1a2b3c4d   ┃
┠───────────────────────────────────────────────────┨
┃ Configuration                                     ┃
┠───────────────────────────────────────────────────┨
┃ enable_cache    │ true                            ┃
┃ retries         │ 3                               ┃
┃ timeout         │ 3600                            ┃
┗━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┛

Delete a Deployment

Remove a deployment:

zenml deployment delete <DEPLOYMENT_NAME_OR_ID>

Example:

zenml deployment delete daily-training

Deleting a deployment stops all scheduled executions and removes the deployment configuration. This action cannot be undone.

Deployment Lifecycle

Creating Deployments

Deployments are typically created using the zenml pipeline deploy command:

zenml pipeline deploy <SOURCE> --name <DEPLOYMENT_NAME> [OPTIONS]

See the pipeline CLI documentation for details. Example:

zenml pipeline deploy my_module.training_pipeline --name daily-training

Deployment Status

Deployments can have these statuses:

Status	Description
`active`	Deployment is running and will execute on schedule/trigger
`inactive`	Deployment is stopped and won’t execute
`failed`	Deployment encountered an error
`starting`	Deployment is being initialized
`stopping`	Deployment is being shut down

Managing Active Deployments

View and filter active deployments:

# List only active deployments
zenml deployment list --status active

# Find deployments that may have issues
zenml deployment list --status failed

# Check specific pipeline's deployments
zenml deployment list --pipeline inference_pipeline --status active

Deployment Configuration

Deployments support various configuration options typically specified in the configuration file used during deployment:

Schedule Configuration

# deploy_config.yaml
schedule:
  cron_expression: "0 2 * * *"  # Daily at 2 AM UTC
  catchup: false  # Don't run missed executions
  start_time: "2024-01-01T00:00:00Z"
  end_time: "2024-12-31T23:59:59Z"

Retry Configuration

# deploy_config.yaml
retries: 3
retry_delay_seconds: 300  # 5 minutes
timeout_seconds: 7200  # 2 hours

Resource Configuration

# deploy_config.yaml
resources:
  cpu_count: 4
  memory: "16GB"
  gpu_count: 1

Common Use Cases

Scheduled Training Pipeline

Deploy a training pipeline to run daily:

# Create schedule configuration
cat > schedule_config.yaml << EOF
schedule:
  cron_expression: "0 2 * * *"  # 2 AM daily
  catchup: false
enable_cache: true
EOF

# Deploy pipeline
zenml pipeline deploy my_module.training_pipeline \
  --name daily-training \
  --config schedule_config.yaml \
  --stack production

# Verify deployment
zenml deployment list --name daily-training

Continuous Inference Service

Deploy an inference pipeline that runs continuously:

zenml pipeline deploy my_module.inference_pipeline \
  --name inference-service \
  --config inference_config.yaml

zenml deployment describe inference-service

Weekly Report Generation

Schedule a weekly reporting pipeline:

cat > weekly_report.yaml << EOF
schedule:
  cron_expression: "0 9 * * 1"  # Monday at 9 AM
EOF

zenml pipeline deploy my_module.report_pipeline \
  --name weekly-report \
  --config weekly_report.yaml

Monitoring Deployments

Check Deployment Health

# List active deployments
zenml deployment list --status active

# Check for failures
zenml deployment list --status failed

# Inspect specific deployment
zenml deployment describe <deployment-name> --show-metadata

View Deployment Runs

See pipeline runs triggered by a deployment:

# List runs for a specific deployment
zenml pipeline runs list --deployment <deployment-name>

# Check recent deployment runs
zenml pipeline runs list \
  --deployment <deployment-name> \
  --sort_by "desc:created" \
  --limit 10

Export Deployment Data

Export deployment information for analysis:

# Export as JSON
zenml deployment list --output json > deployments.json

# Export as CSV
zenml deployment list --output csv > deployments.csv

# Filter and export
zenml deployment list --status active --pipeline training_pipeline --output json

Cron Expression Examples

Common cron schedules for deployments:

Schedule	Cron Expression	Description
Every hour	`0 * * * *`	Top of every hour
Every day at 2 AM	`0 2 * * *`	Daily at 2:00 AM UTC
Every Monday at 9 AM	`0 9 * * 1`	Weekly on Monday morning
Every weekday at 6 PM	`0 18 * * 1-5`	Monday-Friday at 6:00 PM
First day of month	`0 0 1 * *`	Monthly at midnight
Every 15 minutes	`/15 * * *`	Four times per hour
Twice daily	`0 6,18 * * *`	6:00 AM and 6:00 PM

Cron expressions use UTC timezone by default. Convert your local time to UTC when scheduling.

Troubleshooting

Deployment Not Executing

If a deployment isn’t running as expected:

Check status:

zenml deployment describe <deployment-name>

Verify schedule (if applicable):

zenml deployment describe <deployment-name> --show-metadata

Check recent runs:

zenml pipeline runs list --deployment <deployment-name> --limit 5

Verify orchestrator: Ensure your orchestrator supports deployments:
```
zenml stack describe
```

Deployment Failed Status

For failed deployments:

# Get detailed information
zenml deployment describe <deployment-name> --show-metadata

# Check error logs in recent runs
zenml pipeline runs describe <latest-run-id>

# Delete and recreate if needed
zenml deployment delete <deployment-name>
zenml pipeline deploy <source> --name <deployment-name> --config <config>

Updating Deployments

To update a deployment:

# Delete old deployment
zenml deployment delete old-deployment

# Create new deployment with updated configuration
zenml pipeline deploy my_module.pipeline --name new-deployment --config updated_config.yaml

Or use the --update flag when deploying:

zenml pipeline deploy my_module.pipeline --name existing-deployment --config new_config.yaml --update

Best Practices

Naming Conventions

Use descriptive names that indicate purpose and schedule:

daily-training-prod
weekly-batch-inference
hourly-data-refresh
continuous-monitoring

Configuration Management

Keep deployment configurations in version control:

# Store configs in repo
mkdir deployment-configs
cat > deployment-configs/daily-training.yaml << EOF
schedule:
  cron_expression: "0 2 * * *"
enable_cache: true
retries: 3
EOF

# Deploy using versioned config
zenml pipeline deploy my_module.pipeline \
  --name daily-training \
  --config deployment-configs/daily-training.yaml

Monitoring and Alerts

Regularly check deployment health:

# Create monitoring script
cat > check_deployments.sh << 'EOF'
#!/bin/bash
echo "Active Deployments:"
zenml deployment list --status active
echo ""
echo "Failed Deployments:"
zenml deployment list --status failed
EOF

chmod +x check_deployments.sh
./check_deployments.sh

Testing Before Production

Test deployments on staging before production:

# Test on staging
zenml pipeline deploy my_module.pipeline \
  --name test-deployment \
  --stack staging \
  --config test_config.yaml

# Verify it works
zenml deployment describe test-deployment

# Deploy to production
zenml pipeline deploy my_module.pipeline \
  --name prod-deployment \
  --stack production \
  --config prod_config.yaml

Next Steps

Pipeline Commands

Learn about deploying pipelines

Stack Configuration

Configure orchestrators that support deployments

Schedules Guide

Deep dive into scheduling options

Monitoring & Observability

Track deployment performance and health

CLI Commands

zenml deployment

Overview

Command Structure

Core Commands

List Deployments

Describe a Deployment

Delete a Deployment

Deployment Lifecycle

Creating Deployments

Deployment Status

Managing Active Deployments

Deployment Configuration

Schedule Configuration

Retry Configuration

Resource Configuration

Common Use Cases

Scheduled Training Pipeline

Continuous Inference Service

Weekly Report Generation

Monitoring Deployments

Check Deployment Health

View Deployment Runs

Export Deployment Data

Cron Expression Examples

Troubleshooting

Deployment Not Executing

Deployment Failed Status

Updating Deployments

Best Practices

Naming Conventions

Configuration Management

Monitoring and Alerts

Testing Before Production

Next Steps

Pipeline Commands

Stack Configuration

Schedules Guide

Monitoring & Observability

Build docs developers (and LLMs) love

CLI Commands

​Overview

​Command Structure

​Core Commands

​List Deployments

​Describe a Deployment

​Delete a Deployment

​Deployment Lifecycle

​Creating Deployments

​Deployment Status

​Managing Active Deployments

​Deployment Configuration

​Schedule Configuration

​Retry Configuration

​Resource Configuration

​Common Use Cases

​Scheduled Training Pipeline

​Continuous Inference Service

​Weekly Report Generation

​Monitoring Deployments

​Check Deployment Health

​View Deployment Runs

​Export Deployment Data

​Cron Expression Examples

​Troubleshooting

​Deployment Not Executing

​Deployment Failed Status

​Updating Deployments

​Best Practices

​Naming Conventions

​Configuration Management

​Monitoring and Alerts

​Testing Before Production

​Next Steps

Pipeline Commands

Stack Configuration

Schedules Guide

Monitoring & Observability

Build docs developers (and LLMs) love

Overview

Command Structure

Core Commands

List Deployments

Describe a Deployment

Delete a Deployment

Deployment Lifecycle

Creating Deployments

Deployment Status

Managing Active Deployments

Deployment Configuration

Schedule Configuration

Retry Configuration

Resource Configuration

Common Use Cases

Scheduled Training Pipeline

Continuous Inference Service

Weekly Report Generation

Monitoring Deployments

Check Deployment Health

View Deployment Runs

Export Deployment Data

Cron Expression Examples

Troubleshooting

Deployment Not Executing

Deployment Failed Status

Updating Deployments

Best Practices

Naming Conventions

Configuration Management

Monitoring and Alerts

Testing Before Production

Next Steps