This topic provides answers to some frequently asked questions about Alibaba Cloud E-MapReduce (EMR) on ACK. ACK is short for Container Service for Kubernetes.
Why does EMR Shuffle Service fail to start when the number of pods in the ShuffleServiceMaster component is 1?
To ensure high availability and normal running of your cluster, we recommend that you configure at least three pods for the ShuffleServiceMaster component. In the case of multi-replica deployment, EMR Shuffle Service can still remain stable running and serve requests even if errors occur on specific instances. To ensure cluster stability and reliability, you must configure at least three replicas for pods in the ShuffleServiceMaster component.
Three ShuffleServiceWorker components are deployed in an ACK cluster. However, the information 3/4 is displayed in the Number of Pods column for ShuffleServiceWorker on the Service Details tab of the EMR cluster that is associated with the ACK cluster. Why?
In component design, if the number of configured pods does not reach 1,000, one pod is reserved to be in the Pending state. Therefore, if you deploy three ShuffleServiceWorker components in the associated ACK cluster, the information 3/4 is displayed in the Number of Pods column for ShuffleServiceWorker on the Service Details tab of your EMR cluster. This is normal.
What is the meaning of the information 3/1000 that is displayed for shuffleservice-worker on the StatefulSets page of an associated ACK cluster in the ACK console?
The information 3/1000 indicates that the number of ShuffleServiceWorker components that are configured is 3. You can increase the number to 1,000 at most. You can ignore the similar alert information.
0/18 nodes are available: 13 node(s) didn't match Pod's node affinity/selector, 2 node(s) had untolerated taint {ingress-pod: yes}, 3 node(s) didn't match pod anti-affinity rules. preemption: 0/18 nodes are available: 15 Preemption is not helpful for scheduling, 3 No preemption victims found for incoming pod., , preemption: 0/18 nodes are available: 15 Preemption is not helpful for scheduling, 3 No preemption victims found for incoming pod..Can clusters that are created on the EMR on ACK page be used to access different OSS buckets?
No, clusters that are created on the EMR on ACK page cannot be used to access different Object Storage Service (OSS) buckets. Clusters that are created on the EMR on ACK page can be used to read data from or write data to only OSS buckets that they are associated with. If a Spark job is run in a cluster that is created on the EMR on ACK page and is associated with Bucket A and attempts to read data from or write data to Bucket B, the "AccessDenied The bucket you access does not belong to you" error message appears, which indicates that the access is denied.