[YUNIKORN-1246] release reservation of required node in tryAllocate

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.1.0
• Fix Version/s: 1.1.0
• Component/s: core - common
• Labels: pull-request-available
• Target Version: 1.1.0
• Assignee: Wilfred Spiegelenburg
• Reporter: Wilfred Spiegelenburg
• Created: 22/Jun/22 04:22
• Updated: 13/Sep/22 14:31
• Resolved: 28/Jun/22 03:10

Description

When an ask has a required node set we only try to allocate the one specific node. In tryAllocate we have two issues:

• The node specified as the required node might not exist. In that case the scheduler crashes with a nil pointer
• If the release of a preempted allocation is processed during the scheduling cycle we might try to allocate the ask on the node via the normal tryAllocate and not use the tryReservedAllocate flow. If that happens we leave the node reserved forever and that makes the node unusable.

Issue Links

• Discovered while testing: YUNIKORN-1242 gang: delay in scheduling after using all placeholders
• links to: GitHub Pull Request #418