9.2. Disaster Recovery with stretch cluster
In this case, a single cluster is stretched across two zones with a third zone as the location for the arbiter. This feature is currently intended for deployment in the OpenShift Container Platform on-premises and in the same location. This solution is not recommended for deployments stretching over multiple data centers. Instead, consider Metro-DR as a first option for no data loss DR solution deployed over multiple data centers with low latency networks.
The stretch cluster solution is designed for deployments where latencies do not exceed 10 ms maximum round-trip time (RTT) between the zones containing data volumes. For Arbiter nodes follow the latency requirements specified for etcd, see Guidance for Red Hat OpenShift Container Platform Clusters - Deployments Spanning Multiple Sites(Data Centers/Regions). Contact Red Hat Customer Support if you are planning to deploy with higher latencies.
To use the stretch cluster,
You must have a minimum of five nodes across three zones, where:
- Two nodes per zone are used for each data-center zone, and one additional zone with one node is used for arbiter zone (the arbiter can be on a master node).
All the nodes must be manually labeled with the zone labels prior to cluster creation.
For example, the zones can be labeled as:
-
topology.kubernetes.io/zone=arbiter(master or worker node) -
topology.kubernetes.io/zone=datacenter1(minimum two worker nodes) -
topology.kubernetes.io/zone=datacenter2(minimum two worker nodes)
-
For more information, see Configuring OpenShift Data Foundation for stretch cluster.
To know how subscriptions for OpenShift Data Foundation work, see knowledgebase article on OpenShift Data Foundation subscriptions.