Skip to content
This repository has been archived by the owner on Feb 20, 2024. It is now read-only.

Product Deliverables

Mikkin Patel edited this page May 21, 2018 · 2 revisions

The first product deliverable is the following:

  • Updated CP 4.1 Confluent docker images and Helm charts. Covers all Confluent Platform components. 

  • Reference Architecture white paper that describes how to provision Confluent Platform on Kubernetes.

Design considerations to cover:

  • How to expose broker IPs to external clients. Direct client <> broker communication happens over TCP. 

    • Expose the bootstrap address
  • How to configure Zookeeper to know about each Zookeeper node addresses 

  • Docker image best practices:
    • List all ports that are exposed by the Docker images
    • List all volume paths that are needed by the Docker images

Determine recommendations for these environmental considerations:

  • Logging
  • Persistent volumes
  • JMX Monitoring
  • Security

Milestone approach

Dividing up the first product deliverable into milestones:

  • 3 node zookeeper, 3 node Kafka broker, 1 node rest proxy
  • Full Confluent Platform across multiple nodes

Design decisions

  • Logging
    • Confluent Platform does not provide opinions about how to log
    • Options:
      • Kubernetes supports for a bare bones logging
      • (v1) Sidecar approach for gathering and aggregating logs
  • Configuration
    • (v1) Use ConfigMaps - the native way for Kubernetes
    • Passed into Docker images via environment variables, and then a script
    • Consider: Restricting user from editing certain configurations, by providing defaults that make sense for the 90% use cases
    • Look at Zookeeper helm chart
    • Provide a values.yaml to define default configurations passed to template (https://github.com/kubernetes/charts/blob/master/incubator/zookeeper/values.yaml)
  • Security
    • Goal: Be able to deploy Confluent Platform with security enabled on the server side
    • Enable the operator to manually set up
  • Upgrades
    • Rolling upgrades are supported in Confluent Platform
    • Docker images might need to change to accommodate this.
  • Persistent volumes
    • What goes on persistent storage?
      • Data dir logs (logs.dir)
  • Monitoring
    • How do we monitor ZK to ensure that there isn't a split brain scenario? Do we have anything in C3?
  • Scaling
    • What do we need to do to scale down brokers?

May 31st deliverable TODOs:

  1. Configuration parameters for Docker image for Connect.
  2. Monitoring for SR, RP, and Connect
  3. Documentation of umbrella charts
  4. CLA bot
  5. Apache license for cp-helm-chart
  6. Remove wiki entries
Clone this wiki locally