Cross Version

1. Delete the ConfigMap, clean up the Job list and find the ConfigMap configuration, as shown in the following example:

   kubernetes.cluster-id: md-flink
   kubernetes.namespace: default

   Execute the script to batch delete ConfigMap.

   If something like configmap "md-flink" deleted is not output, it means that the namespace or the prefix of ConfigMap is not correct, check again.

   # for i in $(kubectl -n [Replaced with value of kubernetes.namespace] get cm | awk '$1~"[Replaced with value of kubernetes.cluster-id]"{print $1}');do kubectl -n [Replaced with value ofkubernetes.namespace] delete cm $i;done
   
   for i in $(kubectl -n default get cm | awk '$1~"md-flink"{print $1}');do kubectl -n default delete cm $i;done

2. Download the mirror of the new version

   Operations are required on each node server in a kubernetes cluster.

   Internet Access Available

   crictl pull registry.cn-hangzhou.aliyuncs.com/mdpublic/mingdaoyun-flink:<version>

   Internet Access Unavailable

   # Download link for flink offline mirror file. After downloading, upload it to the deployment server.
   https://pdpublic.mingdao.com/private-deployment/offline/mingdaoyun-flink-linux-amd64-<version>.tar.gz

   Load the offline mirror on the server.

   gunzip -d mingdaoyun-flink-linux-amd64-<version>.tar.gz
   ctr -n k8s.io image import mingdaoyun-flink-linux-amd64-<version>.tar

3. Modify the configuration file

   Modify the version of the mirrors used by the flink-jobmanager and flink-taskmanager services in flink. yaml.

   - name: jobmanager
     image: registry.cn-hangzhou.aliyuncs.com/mdpublic/mingdaoyun-flink:<version>
   - name: taskmanager
     image: registry.cn-hangzhou.aliyuncs.com/mdpublic/mingdaoyun-flink:<version>

4. Restart service

   kubectl apply -f flink.yaml

5. Go to [Integrate] > [Data Pipeline] > [Sync Task] to to re-open or publish the task.