[YUNIKORN-528] Nil pointer exception while getting both termination and delete pod event

• Type: Sub-task
• Status: Closed
• Priority: Blocker
• Resolution: Fixed
• Fix Version/s: 0.10
• Component/s: shim - kubernetes
• Labels: pull-request-available
• Assignee: Weiwei Yang
• Reporter: Weiwei Yang
• Created: 30/Jan/21 00:44
• Updated: 21/Jan/22 21:48
• Resolved: 01/Feb/21 08:58

Description

During the test, I observed on some occasions the scheduler could run into Nil pointer exception like below:

4-261f-4448-bc0f-5ea14d23f9e8"}
panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x0 pc=0x190cf5d]

goroutine 114 [running]:
github.com/apache/incubator-yunikorn-core/pkg/scheduler/objects.(*Application).ReplaceAllocation(0xc004250000, 0xc0038e01b0, 0x24, 0x0)
	/Users/wyang/go/pkg/mod/github.com/apache/incubator-yunikorn-core@v0.0.0-20210126213806-78bf4f684709/pkg/scheduler/objects/application.go:1026 +0xcd
github.com/apache/incubator-yunikorn-core/pkg/scheduler.(*PartitionContext).removeAllocation(0xc0026de600, 0xc0003c0a10, 0x0, 0x0, 0x0, 0x0)
	/Users/wyang/go/pkg/mod/github.com/apache/incubator-yunikorn-core@v0.0.0-20210126213806-78bf4f684709/pkg/scheduler/partition.go:1137 +0x14b5
github.com/apache/incubator-yunikorn-core/pkg/scheduler.(*ClusterContext).processAllocationReleases(0xc0001400f0, 0xc0066400c0, 0x1, 0x1, 0x7ffeefbff80f, 0x9)
	/Users/wyang/go/pkg/mod/github.com/apache/incubator-yunikorn-core@v0.0.0-20210126213806-78bf4f684709/pkg/scheduler/context.go:683 +0x150
github.com/apache/incubator-yunikorn-core/pkg/scheduler.(*ClusterContext).processAllocations(0xc0001400f0, 0xc006730000)
	/Users/wyang/go/pkg/mod/github.com/apache/incubator-yunikorn-core@v0.0.0-20210126213806-78bf4f684709/pkg/scheduler/context.go:606 +0x185
github.com/apache/incubator-yunikorn-core/pkg/scheduler.(*ClusterContext).processRMUpdateEvent(0xc0001400f0, 0xc0066ee0b8)
	/Users/wyang/go/pkg/mod/github.com/apache/incubator-yunikorn-core@v0.0.0-20210126213806-78bf4f684709/pkg/scheduler/context.go:213 +0x77
github.com/apache/incubator-yunikorn-core/pkg/scheduler.(*Scheduler).handleRMEvent(0xc00000e3c0)
	/Users/wyang/go/pkg/mod/github.com/apache/incubator-yunikorn-core@v0.0.0-20210126213806-78bf4f684709/pkg/scheduler/scheduler.go:112 +0x416
created by github.com/apache/incubator-yunikorn-core/pkg/scheduler.(*Scheduler).StartService
	/Users/wyang/go/pkg/mod/github.com/apache/incubator-yunikorn-core@v0.0.0-20210126213806-78bf4f684709/pkg/scheduler/scheduler.go:54 +0xa2
make: *** [run] Error 2

the root cause is when the shim deletes a placeholder, it can trigger 2 events sometime,

• Pod Update
• Pod Delete

When a pod updated to TERMINATED state and when a pod gets DELETED, the shim will send a release request to the core. But when there is a second release request, as the previous one already removed the allocation, then we are hitting the Nil pointer. We need to avoid sending a second time release if the pod is already released.

Issue Links

• is caused by: YUNIKORN-229 shim sends the same remove request twice for a remove allocation
• links to: GitHub Pull Request #243