Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

cloudcore crash when no nodeSelector provide #1913

Closed
luogangyi opened this issue Jul 9, 2020 · 6 comments · Fixed by #1916
Closed

cloudcore crash when no nodeSelector provide #1913

luogangyi opened this issue Jul 9, 2020 · 6 comments · Fixed by #1916
Assignees
Labels
kind/bug Categorizes issue or PR as related to a bug.

Comments

@luogangyi
Copy link
Member

What happened:
when we create a device instance without nodeSelector, cloudcore will crash.

panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x0 pc=0x1363d8f]

goroutine 1416 [running]:
github.com/kubeedge/kubeedge/cloud/pkg/devicecontroller/controller.(*DownstreamController).deviceAdded(0xc000562380, 0xc00122e4e0)
        /go/src/github.com/kubeedge/kubeedge/cloud/pkg/devicecontroller/controller/downstream.go:365 +0x9f
github.com/kubeedge/kubeedge/cloud/pkg/devicecontroller/controller.(*DownstreamController).syncDevice(0xc000562380)
        /go/src/github.com/kubeedge/kubeedge/cloud/pkg/devicecontroller/controller/downstream.go:158 +0x24e
created by github.com/kubeedge/kubeedge/cloud/pkg/devicecontroller/controller.(*DownstreamController).Start
        /go/src/github.com/kubeedge/kubeedge/cloud/pkg/devicecontroller/controller/downstream.go:841 +0xd0

What you expected to happen:
cloudcore works properly on device without nodeselector, or just reject it.

How to reproduce it (as minimally and precisely as possible):

[root@masternode1 api-test]# cat create-device-instance-test.yaml
apiVersion: devices.kubeedge.io/v1alpha1
kind: Device
metadata:
  name: sensor-tag-instance-01-test
  labels:
    description: TISimplelinkSensorTag
    manufacturer: TexasInstruments
    model: CC2650
spec:
  deviceModelRef:
    name: sensor-tag-model-test
kubectl create -f create-device-instance-test.yaml

Anything else we need to know?:
I think these codes cause the problem

cloud/pkg/devicecontroller/controller/downstream.go:172:        configMap, ok := dc.configMapManager.ConfigMap.Load(device.Spec.NodeSelector.NodeSelectorTerms[0].MatchExpressions[0].Values[0])
cloud/pkg/devicecontroller/controller/downstream.go:177:                nodeConfigMap.Name = DeviceProfileConfigPrefix + device.Spec.NodeSelector.NodeSelectorTerms[0].MatchExpressions[0].Values[0]
cloud/pkg/devicecontroller/controller/downstream.go:183:                dc.configMapManager.ConfigMap.Store(device.Spec.NodeSelector.NodeSelectorTerms[0].MatchExpressions[0].Values[0], nodeConfigMap)
cloud/pkg/devicecontroller/controller/downstream.go:204:        dc.configMapManager.ConfigMap.Store(device.Spec.NodeSelector.NodeSelectorTerms[0].MatchExpressions[0].Values[0], nodeConfigMap)
cloud/pkg/devicecontroller/controller/downstream.go:365:        if len(device.Spec.NodeSelector.NodeSelectorTerms) != 0 && len(device.Spec.NodeSelector.NodeSelectorTerms[0].MatchExpressions) != 0 && len(device.Spec.NodeSelector.NodeSelectorTerms[0].MatchExpressions[0].Values) != 0 {
cloud/pkg/devicecontroller/controller/downstream.go:370:                resource, err := messagelayer.BuildResource(device.Spec.NodeSelector.NodeSelectorTerms[0].MatchExpressions[0].Values[0], "membership", "")
cloud/pkg/devicecontroller/controller/downstream.go:447:// isNodeSelectorUpdated checks if nodeSelector is updated
cloud/pkg/devicecontroller/controller/downstream.go:448:func isNodeSelectorUpdated(oldTwin *v1.NodeSelector, newTwin *v1.NodeSelector) bool {
cloud/pkg/devicecontroller/controller/downstream.go:449:        return !reflect.DeepEqual(oldTwin.NodeSelectorTerms, newTwin.NodeSelectorTerms)
cloud/pkg/devicecontroller/controller/downstream.go:469:        if len(device.Spec.NodeSelector.NodeSelectorTerms) != 0 && len(device.Spec.NodeSelector.NodeSelectorTerms[0].MatchExpressions) != 0 && len(device.Spec.NodeSelector.NodeSelectorTerms[0].MatchExpressions[0].Values) != 0 {
cloud/pkg/devicecontroller/controller/downstream.go:470:                configMap, ok := dc.configMapManager.ConfigMap.Load(device.Spec.NodeSelector.NodeSelectorTerms[0].MatchExpressions[0].Values[0])
cloud/pkg/devicecontroller/controller/downstream.go:550:                dc.configMapManager.ConfigMap.Store(device.Spec.NodeSelector.NodeSelectorTerms[0].MatchExpressions[0].Values[0], nodeConfigMap)
cloud/pkg/devicecontroller/controller/downstream.go:576:                        if isNodeSelectorUpdated(cachedDevice.Spec.NodeSelector, device.Spec.NodeSelector) {
cloud/pkg/devicecontroller/controller/downstream.go:599:                                        resource, err := messagelayer.BuildResource(device.Spec.NodeSelector.NodeSelectorTerms[0].MatchExpressions[0].Values[0], "device/"+device.Name+"/twin/cloud_updated", "")
cloud/pkg/devicecontroller/controller/downstream.go:697:        if len(device.Spec.NodeSelector.NodeSelectorTerms) != 0 && len(device.Spec.NodeSelector.NodeSelectorTerms[0].MatchExpressions) != 0 && len(device.Spec.NodeSelector.NodeSelectorTerms[0].MatchExpressions[0].Values) != 0 {
cloud/pkg/devicecontroller/controller/downstream.go:698:                configMap, ok := dc.configMapManager.ConfigMap.Load(device.Spec.NodeSelector.NodeSelectorTerms[0].MatchExpressions[0].Values[0])
cloud/pkg/devicecontroller/controller/downstream.go:715:                        dc.configMapManager.ConfigMap.Delete(device.Spec.NodeSelector.NodeSelectorTerms[0].MatchExpressions[0].Values[0])
cloud/pkg/devicecontroller/controller/downstream.go:720:                dc.configMapManager.ConfigMap.Store(device.Spec.NodeSelector.NodeSelectorTerms[0].MatchExpressions[0].Values[0], nodeConfigMap)
cloud/pkg/devicecontroller/controller/downstream.go:811:        if len(device.Spec.NodeSelector.NodeSelectorTerms) != 0 && len(device.Spec.NodeSelector.NodeSelectorTerms[0].MatchExpressions) != 0 && len(device.Spec.NodeSelector.NodeSelectorTerms[0].MatchExpressions[0].Values) != 0 {
cloud/pkg/devicecontroller/controller/downstream.go:812:                resource, err := messagelayer.BuildResource(device.Spec.NodeSelector.NodeSelectorTerms[0].MatchExpressions[0].Values[0], "membership", "")
cloud/pkg/synccontroller/createfailedobject.go:94:              nodeName := device.Spec.NodeSelector.NodeSelectorTerms[0].MatchExpressions[0].Values[0]

Environment:

  • KubeEdge version(e.g. cloudcore/edgecore --version):
    v.1.3.0

CloudSide Environment:

  • Hardware configuration (e.g. lscpu):
  • OS (e.g. cat /etc/os-release):
  • Kernel (e.g. uname -a):
  • Go version (e.g. go version):
  • Others:

EdgeSide Environment:

  • edgecore version (e.g. edgecore --version):
  • Hardware configuration (e.g. lscpu):
  • OS (e.g. cat /etc/os-release):
  • Kernel (e.g. uname -a):
  • Go version (e.g. go version):
  • Others:
@luogangyi luogangyi added the kind/bug Categorizes issue or PR as related to a bug. label Jul 9, 2020
@luogangyi
Copy link
Member Author

/assign

@luogangyi
Copy link
Member Author

/assign

@daixiang0
Copy link
Member

@luogangyi
Copy link
Member Author

Could we set nodeSelector when add device in https://github.com/kubeedge/kubeedge/blob/master/cloud/pkg/devicecontroller/controller/downstream.go#L337?

I did some test, if we set nodeSelector when add device, it's OK on device add and update.
but still crash on device deleted.
It's because we set nodeSelector here is not persistent to etcd.

But, if we want to persist nodeSelector, is it necessary to automatic generate an empty nodeSelector? or just let user do it?

@daixiang0
Copy link
Member

Thinking about existing users, if we add it in definition, is there issue when upgrade?

@luogangyi
Copy link
Member Author

Thinking about existing users, if we add it in definition, is there issue when upgrade?

I think it would be OK for existing user. Becasue they cannot have non-nodeselector device in system so far.
So, if we add it in definition, it will only affect adding new deivce in future

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
kind/bug Categorizes issue or PR as related to a bug.
Projects
None yet
Development

Successfully merging a pull request may close this issue.

2 participants