Khi mà bạn deploy 1 application trên k8s và bạn thấy pod của mình như sau: “is in the cache, so can’t be assumed“
Hiện tại cách của mình là xoá pod đó đi.
Hoặc move pod đó sang 1 node khác
Dưới dây là mình tham khảo 1 post của Trung quốc
Refer to https://www.cnblogs.com/Serverlessops/p/12563692.html , you can know
When kube- scheduler allocates a Pod to a Node, it needs to process the Node’s memory and allocate the Pod to the Node. This process can be called ledger pre-occupation. The pre-occupancy process will mark the state of the Pod as Assumed (in the memory state), and then enter the bind phase, calling kube-apiserver to persist the NodeName of the Pod to etcd. At this time, the state of the Pod is still Assumed. Only when the Pod data has been determined to be allocated to this node through Informer watch will the status be changed to Added, and the Pod scheduling life cycle is roughly like this.
This error indicates that the pod is scheduled to be repeatedly preempted. By viewing the log, it is found that the schedulers of multiple master nodes are scheduled at the same time. When the –config was configured for kube-scheduler, the leaderElect in the configuration file was set to false, resulting in no progress. election. After changing to true, pods can be scheduled normally
版权声明：本文为CSDN博主「含笑饮毒酒_」的原创文章，遵循CC 4.0 BY-SA版权协议，转载请附上原文出处链接及本声明。