Lucidworks recommends installing Fusion without resource limits initially as they can over-complicate the initial setup of your cluster, especially for proof-of-concept / getting started clusters. Resource requests / limits directly impact the number of nodes needed to deploy Fusion. Once your installation is up and running with a critical mass of data, then you can start to fine-tune resource limits for Fusion services.
For production like environments, you should define resource limits to help K8s schedule pods correctly across the nodes in your cluster. This is especially important for K8s clusters that host other namespaces besides Fusion.
If you used the
--with-resource-limits option when running the
./customize_fusion_values.sh script, then you already have resource limits configured for your cluster.
Look for a file named
<provider>_<cluster>_<namespace>_fusion_resources.yaml; if you do not have this file, simply copy
resources.yaml to help you get started with setting the appropriate resource limits.
You can refine the resource requests / limits as you test your cluster’s behavior while preparing to go to production with Fusion.