test case to be improved

[xtcyclist]

  Scenario: Set up slow query at first graph service
    # Set up a slow query which will be killed later.
    Given a graph with space named "nba"
    When executing query via graph 0:
      """
      GO 100000 STEPS FROM "Tim Duncan" OVER like YIELD like._dst
      """
    Then an ExecutionError should be raised at runtime: Execution had been killed

This test case assumed a slow query threshold that this query may exceeded and cause it to be killed. This assumption may not be ture in all cases, and may block normal tck testing procedures.

Better to specifically set a threshold and execute a query that would deterministicly trigger the threshold.

[wey-gu]

cc @zhaojunnana I think we could skip it first.

[wey-gu]

@xtcyclist I thought this case was expecting Scenarios to be run in parallel. Thus the following two Scenarios will kill it (with and without needed permission) but the parallel running of the cases is not always true. ?

[xtcyclist]

@xtcyclist I thought this case was expecting Scenarios to be run in parallel. Thus the following two Scenarios will kill it (with and without needed permission) but the parallel running of the cases is not always true. ?

It's not an issue for CI for the time being. It is blocking my local testing which runs in serial.

We shall not expect all scenarios to be executed in parallel at the first place. If one test case crashed the services, it would be almost impossible to figure out who it was. If we run tck in serial, this case, for example, would block the overall process.

Leave it here for now. I skip it in my own testing environment.

[wey-gu]

True we cannot assume it's always running in parallel. At least we need a label for those cases. And document for devs to add corresponding excluding args when running in serial locally.