Deploy Axon on Kubernetes using the CDM
Use Cloud Deployment Manager to generate a converged Axon deployment for Kubernetes.
Generate WEKA Operator deployment artifacts for an Axon (converged) deployment with the Cloud Deployment Manager (CDM) wizard. Axon co-locates WEKA backends with your workload or GPU nodes and adds a client configuration. Use this deployment type for GPU or compute clusters where storage and workloads share the same nodes.
Access the wizard at Cloud Deployment Manager and sign in with your WEKA account. Select Kubernetes as the target platform, then select Axon (converged).
Before you begin
Complete the same preparation as the Dedicated flow. See Before you begin in Deploy dedicated WEKA on Kubernetes using the CDM.
Procedure
The Axon flow contains 8 steps and reuses the Dedicated flow building blocks.
Complete the following steps as described in Deploy dedicated WEKA on Kubernetes using the CDM: Set up the operator, Define the cluster, Size the capacity, Configure drive signing, and Configure networking. The Protocols step doesn't apply to Axon.
Continue with Configure the client, described below.
Complete Choose driver distribution and Set debuggability as described in the same topic. The CSI driver and Stateless clients steps don't apply to Axon.
Configure the client: The Client step (step 6 of 8) generates a WekaClient resource for the co-located workload or GPU nodes. The client runs on the same nodes as the cluster and uses the cluster network by default.
Set Client Cores: the number of cores allocated to the WEKA client on each node (default: 4).
Keep Advanced Client Network disabled unless the client needs a different network than the cluster. When enabled, set Client Data NICs or Client Device Subnets (CIDR) for the client data path.
Select Next.
The Output manifest includes the generated WekaClient resource, named <cluster>-client, with the new cluster as its target. Apply it together with the rest of the output.
Apply the generated configuration
Apply the generated output the same way as the Dedicated flow. See Apply the generated configuration in Deploy WEKA on Kubernetes using the Cloud Deployment Manager. The output also includes the WekaClient resource generated in the Client step.
To save or reload this configuration, see Save and reload a configuration in Deploy dedicated WEKA on Kubernetes using the CDM.
Related topics
Last updated