Closed songbinliu closed 6 years ago
Here is the content of the move action:
I0907 10:16:36.030798 3249 client_protobuf_endpoint.go:129] [ServerRequestEndpoint][[waitForServerMessage] : received message is: &{ActionRequest:probeType:"OpenShift" accountValue:<key:"targetIdentifier" stringValue:"https://10.10.173.119:8443" > accountValue:<key:"password" stringValue:"****" > accountValue:<key:"username" stringValue:"developer" > actionExecutionDTO:<actionType:MOVE actionItem:<actionType:MOVE uuid:"_COlS0pPXEeeNpdCIShQVzw" targetSE:<entityType:CONTAINER_POD id:"ed86c7fd-73dc-11e7-abf9-00505680effd" displayName:"default/saturated.memory-256-2" commoditiesSold:
in the case of K8s + VCenter, since we already knew the stitching doesn't work well with IP and we chose to use UUID, can we also use uuid for actions?
Yes, we can.
A even better way is to make sure that the entityProperities of VirtualMachine are all sent back. For example, here we can use the property: entityProperties:<namespace:"DEFAULT" name:"KubernetesNodeName" value:"enmaster.octurbo.org" > to get the node name.
But currently, even we have set some entityProperties, some of the properties are not sent back in the actionItem.
This is solved by PR113.
Problem: when kubeturbo executing a move action for Container_POD, it fails with the error log: "Cannot find node with IPs 10.128.0.1".
Env: Openshift Cluster 3.4 OpsMgr 6.0.0, with Market1. Kubeturbo PR93
Possible Reason The newSE field of the ActionItem is a virtual machine: enMaster, whose official IP address is: 10.10.173.119, and Kubernetes uses this IP address to find the host. However, the ActionItem only has an IP of "10.128.0.1".
So this maybe because the vCenter probe fails to get the correct IP address of the virtual machine. (stitching type is "uuid" in kubeturbo, and the stitching has successed).