2 個答案
- 最新
- 最多得票
- 最多評論
0
Thanks, EFS is not an option - only EBS. These applications are actually statefulsets but how does that guarantee the 3 pods will be spread accross each AZ ? Unfortunatly I don't have an EKS cluster to test at the moment.
已回答 5 個月前
0
Topology spread constraints is a good way to ensure that pods does not start in same zone or same host ( based on the constraints definition) However one approach in the usecase could be use to statefulsets[1].
You are right, if you add topology constraint "topologyKey: "topology.kubernetes.io/zone" it will schedule both the pods in different AZs ( based on AZs defined in the cluster)
相關內容
- AWS 官方已更新 1 年前
- AWS 官方已更新 1 年前
Topolgy constraints uses the EKS provided node labels to identify the AZs of the worker nodes and thus schedules the pod on different AZs. Hope this helps in clarifying. https://kubernetes.io/docs/concepts/scheduling-eviction/topology-spread-constraints/