Technology Preview: Hibernating and Resuming ROSA classic Clusters

Updated -

September 2026 Update: Red Hat has made the decision not to proceed with further development of the hibernation and resume feature on ROSA (classic architecture). This feature will not reach general availability on ROSA (classic architecture), and no additional Technology Preview access will be granted for this platform.

Customers looking for cluster hibernation-like behavior or scale-to-zero capabilities are encouraged to move to ROSA (with Hosted Control Planes), which already supports scale-from-zero Machine Pools and will soon support zero worker nodes.

Existing customers who were granted early access to the ROSA (classic architecture) hibernation Technology Preview through a Support Exception may continue to reference the terms and conditions below, which remain in effect for clusters already covered under that exception. No new Support Exceptions for this feature will be issued.

Terms and Conditions

  1. Technology Preview Agreement: By using the hibernation/resume feature, you acknowledge that it falls under the Technology Preview category. A cluster may fail to fully resume to a supportable cluster. Red Hat will not attempt to revive a cluster that fails to resume and you must recreate this cluster.

  2. Non-Production Usage: Ensure that the hibernation/resume feature is tested only on non-production clusters. These clusters should not host critical workloads or data that cannot be recovered.

  3. Limited Support: From the moment a cluster is hibernated until it's fully resumed, it will be under Limited Support as per the ROSA SLA.

  4. OpenShift Version: The cluster should have been initially set up on an OpenShift version 4.5 or later.

  5. Hibernation Duration: A cluster should not be in hibernation for more than 30 consecutive days. It's the user's responsibility to monitor this duration and ensure timely resumption.

  6. Avoiding End-of-Life (EOL): Ensure that the OpenShift version on the hibernated cluster doesn't reach its EOL within 30 days from the start of hibernation. This is to allow ample time for necessary upgrades.

  7. MachineConfigPools Status: Before hibernating, verify that the MachineConfigPools are not updating. Use the command:

    oc get machineconfigpool

  8. Cluster Operators Status: Ensure all cluster operators are in the correct state before hibernation:

    oc get clusteroperators

  9. Hibernation Intervals: After resuming a hibernated cluster, wait for at least 2 hours before initiating another hibernation.

  10. Resumption Time: Be prepared for a wait time of up to 45 minutes for the cluster to become ready after resumption.

  11. AWS Local Zones: Hibernated clusters should not utilize machine pools in AWS Local Zones.

By adhering to these guidelines and terms, you can effectively use the hibernation feature on ROSA classic or OSD clusters while ensuring the safety and integrity of your data and workloads.

Comments