Details
-
Improvement
-
Status: Open
-
Not a Priority
-
Resolution: Unresolved
-
1.11.0
-
None
Description
before 1.11, MiniCluster can be reused in each integration test class. (see TestStreamEnvironment#setAsContext)
In 1.11, after we correct the execution behavior of TableEnvironment, StreamTableEnvironment and BatchTableEnvironment (see FLINK-16363, FLINK-17126), MiniCluster will be created for each test case even in same test class (see org.apache.flink.client.deployment.executors.LocalExecutor). It's better we can reuse MiniCluster like before. One approach is we provide a new kind of MiniCluster factory (such as: SessionMiniClusterFactory) instead of using PerJobMiniClusterFactory. WDYT ?