This tool compares your Kubernetes environment to the SAS Viya platform system requirements. It evaluates a
number of items, such as memory, CPU cores, software versions, and permissions. The output is a web-viewable,
HTML report with the results.
SAS recommends running the tool and resolving any reported issues before beginning a SAS Viya platform deployment in a Kubernetes cluster. The tool cannot account for the dynamic resource allocations that Kubernetes may orchestrate once the SAS Viya platform is deployed. The report and the information therein must be considered a snapshot in time.
The Kubernetes cluster for a SAS Viya platform deployment must meet the requirements documented in SAS® Viya® Platform Operations
Ensure that the Kubernetes version is within the documented range for the selected cloud provider.
If the Kubernetes server version is below the default minimum, a warning will be included in the report.
The tool calculates the aggregate Memory and aggregate vCPUs of your cluster. The aggregate Memory is the sum of the Memory capacity on all the active and running nodes. The aggregate CPU is calculated similarly.
There will be some percentage of Memory on each node in the cluster that is considered "overhead" (consumed by the OS, Kubelet, etc) and therefore not be available for the SAS Viya platform. This tool assumes that 85% of the calculated aggregate Memory is available for the SAS Viya platform, and the memory sizes in the report reflect this.
The calculated aggregate number of vCPUs must equal or exceed the required aggregate number of vCPUs for your deployment offering. The requirements per offering are detailed in the Hardware and Resource Requirements section of the SAS Viya Operations document. The required Memory and vCPUs sizes depend on the instance type used for the node.
Your required aggregates must be specified in the following file:
/viya4-ark/pre_install_report/viya_deployment_settings.ini
Here is an example:
# Total Memory of all worker Nodes in GB. Sum of the Memory on all active node required to deploy a specific offering.
# Set value for required for offering
VIYA_MIN_AGGREGATE_WORKER_MEMORY=448G
# Total CPU of all worker Nodes in millicores. Sum of the vCPUs on all active node required to deploy a specific offering.
# Minimum allowed value = '.001'.
VIYA_MIN_AGGREGATE_WORKER_CPU_CORES=56
If the calculated aggregate memory is less than a percentage (85%) of VIYA_MIN_AGGREGATE_WORKER_MEMORY then the tool will flag a memory issue.
If the calculated aggregate vCPUs is less than VIYA_MIN_AGGREGATE_WORKER_CPU_CORES then the tool will flag a CPU issue.
SAS recommends using the SAS Viya 4 Infrastructure as Code (IaC) tools to create a cluster.
Refer to the following IaC repositories for Microsoft Azure, AWS, GCP or Open Source Kubernetes
For OpenShift refer to the documentation in SAS® Viya® Operations OpenShift
Example: See Sizing Recommendations for a Microsoft Azure deployment under Hardware and Resource Requirements. You can use this information to calculate VIYA_MIN_AGGREGATE_WORKER_MEMORY and VIYA_MIN_AGGREGATE_WORKER_CPU_CORES.
| Offering | CAS Node(s) | System Node | Nodes in User Node Pool(s) |
|---|---|---|---|
| SAS Visual Analytics and SAS Data Preparation | Num of Node: Node 1, CPU: 16, Memory, 128 | Num. of Nodes: 1, CPU: 8, Memory: 64 | Num of Node: 1 per Node User Node Pool, CPU: 8 Memory 64 |
VIYA_MIN_AGGREGATE_WORKER_MEMORY =
CAS Node Memory + System Node Memory + (User Nodes Memory * 4)
VIYA_MIN_AGGREGATE_WORKER_CPU_CORES=
CAS node CPU + System Node CPU + (User Node CPU * 4)
- The tool should be run on a machine from which the Kubernetes command-line interface,
kubectl,
can access the Kubernetes cluster. - The tool requires Python 3.6 or higher.
SAS Viya ARK tools require third-party packages be installed before use. You can install all the required packages by
using the provided requirements.txt file in the following command:
python3 -m pip install -r requirements.txt
Download the latest version of this tool and update required packages with every new software order.
Note: You must set your KUBECONFIG environment variable. KUBECONFIG must have administrator rights to the
cluster where you intend to deploy your SAS Viya platform software.
To obtain a complete report use a KUBECONFIG with administrator rights in the cluster. Otherwise, the report will
not be able to evaluate items such as memory, CPU cores, software versions and other node details. It is not useful
for determining if you are ready to deploy your SAS Viya platform software.
Create the namespace where you plan to deploy SAS Viya platform. A namespace is required to run this tool. Specify the namespace where you plan to deploy SAS Viya platform using the namespace option. If a namespace is not provided, the tool will check the current context for a namespace. Ensure that you are running with the correct namespace.
After obtaining the latest version of this tool, cd to <tool-download-dir>/viya4-ark.
The following command provides usage details:
python3 viya-ark.py pre-install-report -h
The tool generates the pre-install check report, viya_pre_install_report_.html. The report is in a
web-viewable, HTML format.
You can modify the /viya4-ark/pre_install_report/viya_deployment_settings.ini file to alter the
minimum and aggregate settings for CPU and memory on nodes. For more information, see the details in the file.
If you modify the VIYA_K8S_VERSION_MIN to a version less than the minimum Kubernetes version supported by this release of the report tool, you are operating outside the supported capabilities of the report tool. SAS recommends using a release of SAS Viya 4 ARK tools that matches the required minimum you are working with.