Skip to content

LAMBDA.md

Latest commit

 

History

History
89 lines (65 loc) · 2.86 KB

LAMBDA.md

File metadata and controls

89 lines (65 loc) · 2.86 KB
Raw

AWS Lambda Integration

The Awair CLI includes integrated AWS Lambda deployment for scheduled data updates with multi-device support.

Quick Start

# Deploy for a specific device (1-minute intervals)
export AWAIR_TOKEN=your_token_here
AWAIR_DATA_PATH=s3://bucket/awair-17617.parquet \
  awair lambda deploy -s awair-updater-17617 -r 1

# Deploy multiple devices (separate stacks)
AWAIR_DEVICE_ID=17617 AWAIR_DATA_PATH=s3://bucket/awair-17617.parquet \
  awair lambda deploy -s awair-updater-17617 -r 1

AWAIR_DEVICE_ID=137496 AWAIR_DATA_PATH=s3://bucket/awair-137496.parquet \
  awair lambda deploy -s awair-updater-137496 -r 1

# Build package only
awair lambda deploy --dry-run

# Test locally
awair lambda test

# View logs (specify function)
aws logs tail /aws/lambda/awair-updater-17617 --follow

What It Does

Creates a scheduled Lambda function per device that:

  • ✅ Runs every minute via EventBridge (configurable)
  • ✅ Updates device-specific S3 Parquet file with latest sensor data
  • ✅ Uses utz.s3.atomic_edit for safe concurrent updates
  • ✅ Integrates with existing CLI functions (fetch_raw_data, ParquetStorage)
  • ✅ Limited to 1 concurrent execution per function (no race conditions)
  • ✅ Independent scheduling and storage per device

Architecture

Multi-Device Setup:

Device 17617:
  EventBridge (1min) → Lambda (awair-updater-17617) → Awair API → s3://bucket/awair-17617.parquet

Device 137496:
  EventBridge (1min) → Lambda (awair-updater-137496) → Awair API → s3://bucket/awair-137496.parquet

Each Lambda reuses CLI functions (fetch_raw_data, ParquetStorage) for consistency.

Files

  • src/awair/lmbda/app.py - CDK application (infrastructure as code)
  • src/awair/lmbda/deploy.py - CDK deployment script
  • src/awair/lmbda/updater.py - Lambda handler function

CDK Benefits

✅ Type safety: Python classes with IDE completion and validation ✅ Code reuse: Uses existing fetch_raw_data() and ParquetStorage ✅ Consistent behavior: Same data processing as awair raw -r ✅ Easy deployment: Single command from CLI ✅ Unified maintenance: Infrastructure and Lambda code in same project ✅ Better abstractions: Duration.minutes(5) vs "rate(5 minutes)"

Rate Limiting

Per device:

  • 1-minute intervals = 1,440 runs/day
  • Multiple devices: 1,440 × N devices per day
  • Example (2 devices): 2,880 requests/day total
  • Well within limits: Awair Enterprise tier provides sufficient capacity
  • Cost: ~$0.50/month per device (mostly free tier)

Monitoring

# View logs for specific device
aws logs tail /aws/lambda/awair-updater-17617 --follow
aws logs tail /aws/lambda/awair-updater-137496 --follow

# Check function status
aws lambda get-function --function-name awair-updater-17617

# List all awair Lambda functions
aws lambda list-functions | grep awair-updater