[YUNIKORN-6] Deadlock in the scheduler when restoring a node with reservations

• Type: Bug
• Status: Closed
• Priority: Blocker
• Resolution: Fixed
• Fix Version/s: 0.8
• Component/s: core - scheduler
• Labels: pull-request-available
• Target Version: 0.8
• Assignee: Wilfred Spiegelenburg
• Reporter: Wilfred Spiegelenburg
• Created: 03/Mar/20 01:49
• Updated: 21/Jan/22 21:46
• Resolved: 04/Mar/20 23:20

Description

During testing we saw a dead lock in the scheduler that stopped all scheduling.
The scheduler is doing two thing at the time:

• normal scheduling routine
• restoring reservations from a node

The scheduler is compiled by default with race detection and during the same test runs we also saw a data race logged.

Stack and data race log info are attached (cleaned up versions)

Issue Links

• links to: GitHub Pull Request #94