[YUNIKORN-2521] Scheduler deadlock

• Type: Bug
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: 1.5.0
• Fix Version/s: 1.6.0, 1.5.1
• Environment: Yunikorn: 1.5 ; AWS EKS: v1.28.6-eks-508b6b3
• Target Version: 1.6.0, 1.5.1
• Assignee: Peter Bacsko
• Reporter: Noah Yoshida
• Created: 28/Mar/24 19:38
• Updated: 23/May/24 03:54
• Resolved: 23/Apr/24 19:42

Description

Discussion on Yunikorn slack: https://yunikornworkspace.slack.com/archives/CLNUW68MU/p1711048995187179

Occasionally, Yunikorn will deadlock and prevent any new pods from starting. All pods stay in Pending. There are no error logs inside of the Yunikorn scheduler indicating any issue. 

Additionally, the pods all have the correct annotations / labels from the admission service, so they are at least getting put into k8s correctly. 

The issue was seen intermittently on Yunikorn version 1.5 in EKS, using version `v1.28.6-eks-508b6b3`. 

At least for me, we run about 25-50 nodes and 200-400 pods. Pods and nodes are added and removed pretty frequently as we do ML workloads. 

Attached is the goroutine dump. We were not able to get a statedump as the endpoint kept timing out. 

You can fix it by restarting the Yunikorn scheduler pod. Sometimes you also have to delete any "Pending" pods that got stuck while the scheduler was deadlocked as well, for them to get picked up by the new scheduler pod. 

Issue Links

• is related to: YUNIKORN-2629 Adding a node can result in a deadlock
• relates to: YUNIKORN-2539 Add optional deadlock detection
• relates to: YUNIKORN-2544 [UMBRELLA] Fix Yunikorn potential locking issues