Describe the feature:
We’d like to have more robust automated test coverage for the Upgrade Assistant. In order to do so, we need the ability to run multi-version clusters, i.e., do X in old version, upgrade cluster, then do Y in new version. Elasticsearch offers similar capabilities with its backwards compatibility tests.
Describe a specific use case for the feature:
For the 8.0 Upgrade Assistant, we are supporting upgrading or deleting old Machine Learning snapshots. Currently all tests are written using mocked data (related issue: https://github.com/elastic/kibana/issues/91879).
It’d be great if we could support integration tests where a 6.x cluster could be spun up that would create the old ML snapshots, then upgraded to 7.x. On the 7.x cluster, we can verify that the deprecation is triggered and the Upgrade Assistant successfully resolves it.
Describe the feature: We’d like to have more robust automated test coverage for the Upgrade Assistant. In order to do so, we need the ability to run multi-version clusters, i.e., do X in old version, upgrade cluster, then do Y in new version. Elasticsearch offers similar capabilities with its backwards compatibility tests.
Describe a specific use case for the feature: For the 8.0 Upgrade Assistant, we are supporting upgrading or deleting old Machine Learning snapshots. Currently all tests are written using mocked data (related issue: https://github.com/elastic/kibana/issues/91879).
It’d be great if we could support integration tests where a 6.x cluster could be spun up that would create the old ML snapshots, then upgraded to 7.x. On the 7.x cluster, we can verify that the deprecation is triggered and the Upgrade Assistant successfully resolves it.