Skip to end of metadata
Go to start of metadata

You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 15 Next »

Goals

  • CDAP and CDAP Applications have the ability to withstand short and transient infrastructural outages
  • During interruption of underlying services (one or more), CDAP or CDAP Applications can operate under degraded performance/limited functionalities 
    • Users will not be able to perform operations like deploying apps, starting programs or new data or application lifecycle operations.
      • However all the applications that are running, should be running
  • Once interruption in the underlying service is resolved or services come back to normal operation, the CDAP and CDAP Application will go back to normal state 
  • Interruptions in service would be due to node failure, service failures or compatible rolling upgrades or downgrades in progress
  • Does not include in-compatible upgrades or downgrades of underlying infrastructure 
  • Does not include long unavailability of service and infrastructure

Area of Focus

  • CDAP system resiliency to infrastructure unavailability or interruption for long periods of time
  • CDAP Rolling Upgrades
  • CDAP Application Rolling Upgrades

High Level Requirements

  • Compatible Upgrade or Downgrade of underlying infrastructure Hadoop components
    • Underlying Hadoop infrastructure is either being upgraded or downgraded and the expectation is that CDAP and CDAP Applications should tolerate and be resilient to infrastructure services not being available during the upgrade or downgrade process. 
    • The upgrade or downgrade process could take anywhere between 30 mins - 18 hours or more. 
    • During the period of service unavailability or interruption, the CDAP and CDAP Applications operate in degraded mode.
    • Hadoop infrastructure upgrade / downgrade has to be compatible with CDAP and CDAP Application in order to have smooth upgrades
    • In case, there are issues during the upgrade, CDAP should be resilient to rollbacks
    • CDAP and CDAP Applications should also be able to withstand compatible downgrades
    • The compatibility matrix should be available to users to ensure smooth upgrades
  • Upgrade / Downgrade of CDAP
    • Upgrade a CDAP version. Major and minor version could have different impacts. We will discuss about these impact further in the document. 
    • Roll back of CDAP upgrade or downgrade
    • CDAP version compatible matrix available to users
    • Dry run for upgrade and downgrade
  • Upgrade / Downgrade of CDAP Applications
    • Upgrade or Downgrade a CDAP Application

    • Rolling upgrade of live services like CDAP Services, Flow and Spark Streaming

Technical Breakdown

RS-001 : Un-interrupt update of compact modules in Coprocessor

CDAP system uses few HBase coprocessors to optimize the operations being performed on HBase. When underlying HBase is upgraded, the table has to be altered. This means that the table has to be disabled. Disabling the table can have multiple side effects on CDAP, so the recommended approach right now is to stop applications running within CDAP as well as CDAP. For each version of non-compatible HBase, CDAP has a compat module has to updated.

RS-002 : Client Resiliency

RS-003 : Move Dataset management out of CDAP Master

RS-004 : CDAP Version definition and guarantees of versions

RS-005 : Rolling upgrade definition

 RS-006 : Internal Schema Evolution and Management

RS-007 : Managing Infrastructure Incompatibility 

RS-008 : System state transition and management 

RS-009 : Apache Twill Application rolling upgrade 

RS-010 : 

  • HIGH Client Resiliency

  • Rolling Upgrade Definition
  • Internal Schema Evolution
  • Infrastructure Incompatibility 
  • State Transition and Management
  • Apache Twill Rolling Upgrade Support
  • Rolling upgrade orchestrator
  • Progressive background upgrade tool
  • User Interface / REST APIs / CLI
  • Testing Framework and Chaos monkey
  • Hydrator pipeline upgradability

Open Item/Discussion point

  • Define long and short/transient outages
    • More information to gathered here to understand the length of outages. 
    • When outages are multiple hours, how should the system handle these. 

Action Items - Oct 7th 2016

  • (tick) Send supported HBase version by CDAP
  • Gather information about CDH version compatibility changes – Talk to Cloudera and compile 

Initiatives In Progress

  • [3.6] CDAP Service version and upgrade support
  • [3.6] Application versioning
  • [4.0] Messaging Service with goal of centralizing all transactional activities for metadata in HBase
  • [4.0?] Non-Transactional datasets
  • [4.0?] HBase Coprocessor Upgrade Management — Handling minor version changes efficiently without disabling HBase Tables. 
  • [4.0?] Upgrade tool improvements — Coprocessor Upgrade removal, faster data conversions if needed, smarts to reduce the impact to running services
  • [4.0?] CDAP Service Upgrade capability, might have Apache Twill change
  • [4.0?] Move configuration and operational updates to messaging services

Initiatives In Plan

  • Clients have retry and back-off mechanism to operate in degraded mode
  • YARN application resilience through Apache Twill
  • Move Dataset Service that currently runs in Master as YARN Application
  • No labels