Skip to content

Latest commit

 

History

History
17 lines (14 loc) · 782 Bytes

File metadata and controls

17 lines (14 loc) · 782 Bytes

Deployment

Recommended production topology:

  • Flink cluster (Kubernetes via Helm)
  • Flink Job Gateway as a Deployment/Service
  • Kafka as source/sink (managed or self-hosted)

Key notes:

  • Build and publish the Gateway; it prebuilds and bundles flink-ir-runner.jar automatically.
  • The Gateway uploads the runner jar to Flink on submission; no separate jar mounting step is required.
  • Configure Gateway with FLINK_CLUSTER_HOST and FLINK_CLUSTER_PORT.
  • Use parallelism in JobMetadata to scale jobs appropriately.

Security & reliability:

  • Lock down Gateway with auth (API keys, mTLS) in front of cluster network.
  • Enable Flink checkpoints and state backends suitable for durability.
  • Add retries and idempotency around submissions (consider Temporal orchestration).