[IGNITE-7648] Fix IGNITE_ENABLE_FORCIBLE_NODE_KILL system property.

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.3
• Fix Version/s: 2.8
• Assignee: Pavel Voronkin
• Reporter: Alexey Scherbakov
• Created: 08/Feb/18 09:13
• Updated: 10/Feb/19 11:50
• Resolved: 07/Feb/19 12:25

Description

IGNITE_ENABLE_FORCIBLE_NODE_KILL system property was introduced in IGNITE-5718 as a way to prevent unnecessary node drops in case of short network problems.

I suppose it's wrong decision to fix it in such way.

We had faced some issues in our production due to lack of automatic kicking of ill-behaving nodes (on example, hanging due to long GC pauses) until we realised the necessity of changing default behavior via property.

Right solution is to kick nodes only if failure threshold is reached. Such behavior should be always enabled.

UPDATE: During a discussion it was decided what the property will remain disabled by default.

We decided to change timeout logic in case of failure detection enabled. We start performing connect and handshake from 500ms increasing using exponential backoff strategy.

 

 

Issue Links

• duplicates: IGNITE-11281 Cache (Restarts) 1 flaky tests SEGMENTATION problem
• is caused by: IGNITE-5718 Forcible node drop makes cluster instable in some cases.
• is related to: IGNITE-11221 Refactor timeout logic in TcpDiscovery
• is related to: IGNITE-11255 Fix test failure after IGNITE-7648.
• is related to: IGNITE-11126 Rework TcpCommunicationSpi.createShmemClient failure detection logic.
• links to: PR
• links to: TC
• links to: upsource