[IGNITE-11147] Re-balance cancellation occur by non-affected event

• Type: Test
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: 2.7
• Fix Version/s: 2.9
• Component/s: cache
• Ignite Flags: Docs Required
• Assignee: Vladislav Pyatkov
• Reporter: Sergey Antonov
• Reviewer: Alexey Scherbakov
• Created: 30/Jan/19 15:15
• Updated: 06/Aug/21 14:40
• Resolved: 15/May/20 10:46

Description

Re-balance cancels by non-affected events, for examples:
1) joining non affinity node
2) starting snapshot
3) starting/stopping other cache

Try to skip as much as possible events instead of cancellation.

After solved several issues appearing during this testing, I decided to add specific property allowing on/off rebalance's optimization. (see IgniteSystemProperties#IGNITE_DISABLE_REBALANCING_CANCELLATION_OPTIMIZATION by default false).

Issue Links

• is related to: IGNITE-10078 Node failure during concurrent partition updates may cause partition desync between primary and backup.
• relates to: IGNITE-12689 Partitions should become owned after a checkpoint, regardless of a topology change. Nevertheless a rebalance is not required.
• relates to: IGNITE-12290 Re-balance fully restart for case when WAL disabled
• links to: GitHub Pull Request #6489
• links to: GitHub Pull Request #7428