During 2.15.0 release we changed how integration tests are run for OSD by splitting them into multiple ci-groups and executing them separately with each group having its own cluster. This enabled us to have 100% tests passing.
As of now FT repo follows the same logic that OSD uses for grouping and infra team relies on FT repo to decide the number of groups it is supposed to run integration tests for.
There is no mechanism right now to synchronize the grouping logic between OSD, FT Repo and opensearch-build repo so that in case OSD makes any changes to the logic on their side it is reflected everywhere.
Creating this issue to get ideas on how we can have a single source of truth for grouping logic. This will also be used to track ci-groups required for 2.16 release.
During 2.15.0 release we changed how integration tests are run for OSD by splitting them into multiple ci-groups and executing them separately with each group having its own cluster. This enabled us to have 100% tests passing.
As of now FT repo follows the same logic that OSD uses for grouping and infra team relies on FT repo to decide the number of groups it is supposed to run integration tests for. There is no mechanism right now to synchronize the grouping logic between OSD, FT Repo and opensearch-build repo so that in case OSD makes any changes to the logic on their side it is reflected everywhere.
Creating this issue to get ideas on how we can have a single source of truth for grouping logic. This will also be used to track ci-groups required for 2.16 release.
https://github.com/opensearch-project/opensearch-dashboards-functional-test/pull/1411 https://github.com/opensearch-project/opensearch-build/issues/4745