A user attempts to import a large FAR filean archive file that has been exported from a SAS Forecast Server project. The file contains the data definitions, properties, events, and other resources needed to run the project in another environment. The other environment could be a SAS Forecast Server or a SAS Visual Forecasting deployment.. The project is not created.
The SAS administrator can check the job in SAS Environment Manager. When the project is being imported, the Name column should show an import job listing like the following:
Importing far file with task id 133057873
If the Status column shows that the import job has timed out, the FAR file might have been too large to import. It is likely that the sas-forecasting-gateway service was terminated. The SAS administrator can issue this command to make that determination.
kubectl get pods -o json | jq '.items[] | .metadata.name as $name |
.status.containerStatuses[]?.lastState.terminated? | select( .reason == "OOMKilled")
| {($name): (.)}'
The following response shows that the sas-forecasting-gateway was terminated.
"sas-forecasting-gateway-dd84f8bb7-79kk9": {
"containerID": "containerd://b20bfd4a359cc307af7e7d263c2804d8b75fe50af2e4dd9
d9f459c111e22fc3c",
"exitCode": 1,
"finishedAt": "2024-03-08T02:24:04Z",
"reason": "OOMKilled",
"startedAt": "2024-03-08T02:22:26Z"
}
The OOMKilled reason indicates that the POD was terminated because the import needed more memory than the allocated limit.
To request more memory, issue this command to edit the sas-forecasting-gateway pod.
kubectl edit deployment sas-forecasting-gateway
This opens the JSON configuration for the sas-forecasting-gateway service in an editor. Look for these lines:
resources:
limits:
cpu: 500m
memory: 1000Mi
requests:
cpu: 50m
memory: 50Mi
Update the limit to more memory (for example, 5000Mi). Save and close the file.