[FLINK-24184] Potential race condition leading to incorrectly issued interruptions

• Type: Bug
• Status: Closed
• Priority: Not a Priority
• Resolution: Fixed
• Affects Version/s: 1.8.3, 1.9.3, 1.10.3, 1.11.4, 1.14.0, 1.12.5, 1.13.2
• Fix Version/s: 1.14.0, 1.15.0
• Component/s: Runtime / Task
• Labels: pull-request-available
• Assignee: Piotr Nowojski
• Reporter: Piotr Nowojski
• Created: 07/Sep/21 09:04
• Updated: 13/Sep/21 07:28
• Resolved: 13/Sep/21 07:27

Description

There is a race condition in disabling interrupts while closing resources. Currently this is guarded by a volatile variable, but there might be a race condition when:
1. interrupter thread first checked the shouldInterruptOnCancel flag
2. shouldInterruptOnCancel flag switched to false as Task/StreamTask entered cleaning up phase
3. interrupter issued an interrupt while Task/StreamTask are closing/releasing resources, potentially causing a memory leak

Issue Links

• is caused by: FLINK-9776 Interrupt TaskThread only while in User/Operator code
• relates to: FLINK-24182 Tasks canceler should not immediately interrupt
• supercedes: FLINK-21821 Clean up Task/StreamTask cancellation and remove unnecessary concurrency
• links to: GitHub Pull Request #17218