Akka Sprint Plan 2017-06-26

[patriknw]

Sprint plan for the core Akka team

3 weeks Start: 2017-06-26 End: 2017-07-14

Data center aware cluster membership

• Membership, leaders
• Failure detection
• Split Brain Resolver
• Singleton
• Sharding

Other

• Complete hardening of Artery lanes > 1, and performance optimization by implementing PartitionHub
• Akka HTTP Quickstart guide
• Run performance and scalability tests on Google Cloud if we get access to the full scale environment
• More requirements gathering of other multi-DC features
• Help / unblock Play team if they need help after the 2.6 release

Bugs and failures

Bug start count: 12 Failure start count: 38

Akka HTTP Bug start count: 38 Akka HTTP Failure start count: 12

[ktoso]

Checklist for things to finish this week here;

• [x] Complete remaining tasks and final polish of the multi-dc stuff, perhaps some additional tests? https://github.com/akka/akka/issues?q=is%3Aopen+is%3Aissue+label%3At%3Acluster%3Adc
• [x] Review Getting Started PR https://github.com/akka/akka/pull/23210
• [x] Final review and merge of the akka.io PR https://github.com/akka/akka.github.com/pull/451
• [ ] Review HTTP Quickstart PR, and involve Ruth https://github.com/akka/akka-http-scala-seed.g8/pull/11
• [x] Publish new sharding example to TechHub

[ktoso]

Data center aware cluster membership

// Note: "done" here means "for this sprint", by definition we will re-visit those after more distributed testing and continue covering edge cases. These features are not yet ready to be released.

• Membership, leaders - done
  • DatacenterReachable events - we have the events done
  • Not passing new tests cross DC tests yet, could be failure detector issues?
  • TODO: Rethink when to fire the event - when all nodes are split off or just a few
• Failure detection - done, though needs more testing
• Split Brain Resolver - worked on adjusting it to multi DC settings; Is not aimed to cover automatic downing of DCs, that's not the usual use case
• Singleton - done
  • works within one DC, we documented how to do cross DC if really needed
  • docs https://github.com/akka/akka/blob/wip-multi-dc-dev/akka-docs/src/main/paradox/scala/cluster-dc.md#cluster-singleton
• Sharding - done
  • works within one DC, for cross DC we'll have a separate feature
  • docs https://github.com/akka/akka/blob/wip-multi-dc-dev/akka-docs/src/main/paradox/scala/cluster-dc.md#cluster-sharding
• Additional: worked on fixes for DData

Other

• Complete hardening of Artery lanes > 1, and performance optimization by implementing PartitionHub - done, final review and merge https://github.com/akka/akka/pull/23102
• Akka HTTP Quickstart guide
  • Content is there, polishing/editing still in progress, will not hinder us in next sprint; it's going on async
• Run performance and scalability tests on Google Cloud if we get access to the full scale environment
  • Slipped waiting for environment to be provided
• Gathering more requirements of other multi-DC features - done and in progress still
• Help / unblock Play team if they need help after the 2.6 release
  • We’ll continue on this, more actively this sprint

Bugs and failures

• Bug start count: 12

  • Bug end count: 16 (closed 513)

• Failure start count: 38

  • Failure end count: 44 (closed 373) -- many new in the branch about multi-dc

• Akka HTTP Bug start count: 38

  • Akka HTTP Bug end count: 42 (closed 83)

• Akka HTTP Failure start count: 12

  • Akka HTTP Failure end count: 13 (closed 25)

[patriknw]

Have you created issues for remaining dc work so that we know what is left? E.g. What is missing for sharding?

[ktoso]

Not yet, will do though. Sharding mostly you researched if I remember correctly, so we weren't as sure what's missing.

[patriknw]

No, I implemented it.

[ktoso]

Team memory about it was a bit vague :-) Will go over issues and open things in my morning :)

[ktoso]

Fixed status of singleton / sharding, opened follow up tickets