Build Topology with only required levels (FaultZone and EndNode)

[zpinto]

Issues

• Change all rebalancer strategies to create Topology without additional non-FaultZone or EndNode levels of the tree. This will allow for swap to work in clusters where the non-FaultZone or EndNode domain kv pairs don't directly match the swapping node. Helix Node/Instance Swap #2662

Description

• When attempting to use Swap in clusters that had more complex topologies with levels that were not fault zone or end node, it was discovered that Swap assignment was being computed correctly. This is because the levels and all nodes in the Topology tree affect the assignment in CRUSH and CRUSHED. This changes the behavior of those rebalancer strategies to only create levels in the Topology tree for FaultZone and EndNode. This is the only thing required as the topology aware placement should only be done based on FaultZone.

Tests

• Updated TestInstanceOperation to reproduce issues seen in testing on production like cluster. After topology changes, those tests pass.
• Used current changes to deploy local controller for CRUSHED and WAGED clusters copied from production. Successfully carried out SWAP operation for both.

Changes that Break Backward Compatibility (Optional)

Changes to Topology class are not backwards compatible; however, the rebalancer will have different behavior and produce new assignment for existing clusters using more than a 2-layer topology. This constitutes bumping minor version and calling out in release notes in next open source release.

Documentation (Optional)

Commits

• My commits all reference appropriate Apache Helix GitHub issues in their subject lines. In addition, my commits follow the guidelines from "How to write a good git commit message":
  1. Subject is separated from body by a blank line
  2. Subject is limited to 50 characters (not including Jira issue reference)
  3. Subject does not end with a period
  4. Subject uses the imperative mood ("add", not "adding")
  5. Body wraps at 72 characters
  6. Body explains "what" and "why", not "how"

Code Quality

• My diff has been formatted using helix-style.xml
  (helix-style-intellij.xml if IntelliJ IDE is used)

[xyuanlu]

Just curious, Please correct me if I cam wrong, I did not find anywhere we are using the default constructor. We always set faultZoneLevelOnly to true?

[zpinto]

Yes, that is correct; however, this constructor is public and may be used outside our code base by others.

This will keep backwards compatibility, while only changing the behavior of the rebalancer to limit the backwards incompatibility concerns.

[xyuanlu]

Yes, we should always keep the default constructor.

[zpinto]

Failed CI due to flaky test testDisconnectWhenConnectionBreak #2691

This PR is ready to be merged, thanks for the review.

Final commit message:
Change all rebalancer strategies to create Topology without additional non-FaultZone or EndNode levels of the tree.