Stream Compute Service
- Releases Notes and Announcements
- Getting Started
- Operation Guide
- Managing Jobs
- Developing Jobs
- Monitoring Jobs
- Events and Diagnosis
- Managing Metadata
- Tuning Jobs
- Managing Clusters
- Managing Permissions
- SQL Developer Guide
- DDL Statements
- Connectors
- SET Statement
- Operators and Built-in Functions
- Identifiers and Reserved Words
- ETL Developer Guide
Migrating a Cluster
Last updated: 2023-11-07 15:48:54
When your cluster is at a lower edition, some new features (such as metric monitoring and checkpointing) may be unavailable to it. The optimal solution is to migrate your jobs to a cluster at a higher edition.
This document describes how to migrate a cluster in the Stream Compute Service console. If you have any trouble during the migration, submit a ticket to contact us.
Creating a new cluster
Log in to the Stream Compute Service console, and click Create in Compute resources. For details, see Creating a Private Cluster.
Note
When you create a new private cluster, the region selected must be the same as that of the VPC and the existing cluster to be migrated.
Custom configurations of the existing cluster (such as DNS configurations) need to be synced to the new cluster.
The new cluster shall be associated with the workspace of the existing cluster.
Initializing jobs on the existing cluster to the new cluster
The jobs on the existing cluster can be copied to the new cluster via the "batch create copy" feature detailed in Job Operations.
The steps are as follows: Go to the workspace associated with the existing cluster, click Batch on the top of the job list, and select Create copy.
Note
If the destination cluster does not support the Flink version used by the source jobs, the job copies will use the default Flink version of the destination cluster.
If the source jobs use fine-grained resources that are not supported by the destination cluster, both the JobManager spec and TaskManager spec of the job copies will be set to 1 CU.
Jobs can be copied only within the workspace of the existing cluster, so the destination cluster selected must be one associated with the workspace.
Starting/Stopping the jobs on the new/existing cluster
You need to stop jobs on the existing cluster, and start those on the new cluster. After the jobs on the new cluster are started, check the job status for exceptions.
Stop the jobs on the existing cluster
Start the jobs on the new cluster
Note
New jobs will not retain the state data of the original jobs. Please separately consider the data completion logic.
Terminating the existing cluster
After all jobs are successfully migrated, terminate the existing cluster as instructed in Terminating a Cluster.
Was this page helpful?
You can also Contact Sales or Submit a Ticket for help.
Yes
No