Platform detection/validation

[mbukatov]

When I execute tests on a cluster installed outside of ocs-ci, so that I have a standard cluster dir (from openshift-installer) only, like this:

$ run-ci --cluster-path cluster-2020-06-19.1 tests/manage/monitoring/prometheusmetrics/test_monitoring_defaults.py -vvv

I see that ocs-ci assumes that the platform is AWS IPI:

2020-06-19 19:39:39,964 - MainThread - INFO - tests.conftest.cluster.1043 - All logs located at /tmp/ocs-ci-logs-1592588358
2020-06-19 19:39:39,967 - MainThread - INFO - ocs_ci.deployment.factory.get_deployment.43 - Deployment key = aws_ipi
2020-06-19 19:39:39,968 - MainThread - INFO - ocs_ci.deployment.factory.get_deployment.45 - Current deployment platform: AWS,deployment type: ipi

Even though the platform is Azure, which could be seen in metadata.json file created by openshift-installer in the cluster dir:

$ jq '.' cluster-2020-06-19.1/metadata.json 
{
  "clusterName": "mbukatov-2020-06-19-1",
  "clusterID": "0f76c648-a12c-44f2-8ae0-95502d81e27a",
  "infraID": "mbukatov-2020-06-19-1-vm9gx",
  "azure": {
    "region": "eastus"
  }
}

While we can't be sure about IPI vs UPI status (and other deployment options), we could at least detect that the assumptions we have (coming from the default config) doesn't hold, and raise a warning (along with a suggestion what to do) stopping the test.

This could help people who are trying to run tests on clusters deployed outside of our CI, but tries to share results with us.

[mbukatov]

One could try to resolve this by creating a config file with platform info, eg.:

$ cat azure.yaml                                                                
---                                                                             
ENV_DATA:                                                                       
  platform: 'azure'                                                             
  deployment_type: 'ipi

passing it to run-ci via --ocsci-conf, eg:

$ run-ci --cluster-path cluster-2020-06-19.1 --ocsci-conf  azure.yaml tests/manage/monitoring/prometheusmetrics/test_monitoring_defaults.py

[vasukulkarni]

I think solving this will solve many other issues with config file loading for different platforms. It can detect and overwrite the default ones.

[Avilir]

there is function : node.get_provider() which return the actual Provider name that can be use instead of getting t from config file.

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed in 30 days if no further activity occurs.

[github-actions]

This issue has been automatically closed due to inactivity. Please re-open if this still requires investigation.