[IGNITE-9913] Prevent data updates blocking in case of backup BLT server node leave

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 2.8, 2.9
• Component/s: general
• Labels: iep-45
• Release Note: Merged to master branch.
• Ignite Flags: Release Notes Required
• Assignee: Anton Vinogradov (Obsolete, actual is "av")
• Reporter: Ivan Rakov
• Created: 17/Oct/18 13:18
• Updated: 30/Apr/20 09:22
• Resolved: 19/Dec/19 10:09

Description

Ignite cluster performs distributed partition map exchange when any server node leaves or joins the topology.
Distributed PME blocks all updates and may take a long time. If all partitions are assigned according to the baseline topology and server node leaves, there's no actual need to perform distributed PME: every cluster node is able to recalculate new affinity assigments and partition states locally. If we'll implement such lightweight PME and handle mapping and lock requests on new topology version correctly, updates won't be stopped (except updates of partitions that lost their primary copy).

Issue Links

• is related to: IGNITE-10284 performance drop on PME time during 1 server node left
• is related to: IGNITE-11936 Avoid changing AffinityTopologyVersion on a server node join/left event from not baseline topology.
• is related to: IGNITE-12741 Allow exchange merges for PME free switch.
• is related to: IGNITE-12273 Slow TX recovery
• is related to: IGNITE-12470 Pme-free switch feature should be deactivatable
• is related to: IGNITE-12272 Delayed TX recovery
• is related to: IGNITE-12617 PME-free switch should wait for recovery only at affected nodes.
• is related to: IGNITE-12788 Cluster achieved fully rebalanced (PME-free ready) state metric
• links to: GitHub Pull Request #6247
• links to: GitHub Pull Request #6942
• links to: GitHub Pull Request #7069
• links to: GitHub Pull Request #7165
• links to: Non-blocking-PME-Phase-One-Node-fail