[IGNITE-13465] Ignite cluster falls apart if two nodes segmented sequentially

• Type: Bug
• Status: Resolved
• Priority: Blocker
• Resolution: Fixed
• Affects Version/s: 2.9
• Fix Version/s: 2.9.1
• Release Note: Fixed processing of connection recovery timeout in TcpDiscoverySpi. If node loses connection, now it strictly obtains new connection to the ring or becomes segmented within this timeout.
• Assignee: Vladimir Steshin
• Reporter: Aleksey Plekhanov
• Created: 21/Sep/20 11:46
• Updated: 05/Aug/21 11:56
• Resolved: 26/Oct/20 06:58

Description

After [1] sequential nodes failure leads to segmentation of other nodes in the cluster. Reproducer attached.

This ticked contains:

• Reverted [1].
• Fix for [1] to avoid the noted segmenttion.
• Reverted [2] which depends on [1].

[1] IGNITE-13134
[2] IGNITE-13208

Issue Links

• contains: IGNITE-13134 Fix connection recovery timeout.
• contains: IGNITE-13208 Simplify IgniteSpiOperationTimeoutHelper
• links to: GitHub Pull Request #8262
• links to: GitHub Pull Request #8264
• links to: GitHub Pull Request #8265