Dmitrii, The solution is not clear to me. In case you expect the failure then a correct case is to wrap it with try-catch block instead of no-op failure handler usage.
вт, 4 дек. 2018 г. в 21:41, Dmitrii Ryabov <somefire...@gmail.com>: > Anton, > > Tests in these classes check fail cases when we expect critical > failure like node stop or exception thrown. Such tests trigger failure > handler and it fails test when everything goes as it should go. That's > why we need no-op handler here. > вт, 4 дек. 2018 г. в 20:06, Dmitriy Pavlov <dpav...@apache.org>: > > > > Hi Igniters, > > > > BTW, if you find in any of your tests it does't need an old value of > > handler (=NoOp), feel free to remove it. > > > > Sincerely, > > Dmitriy Pavlov > > > > вт, 4 дек. 2018 г. в 20:02, Anton Vinogradov <a...@apache.org>: > > > > > Dmitrii, > > > > > > Could you please explain the reason of explicit set of 100+ > > > NoOpFailureHandlers? > > > > > > > > > вт, 4 дек. 2018 г. в 19:12, Dmitrii Ryabov <somefire...@gmail.com>: > > > > > > > Hello, Igniters! > > > > > > > > Today the test framework's default no-op failure handler was changed > to > > > the > > > > handler, which stops the node and fails the test. > > > > > > > > Over 100 tests kept no-op failure handler by overrided > > > > `getFailureHandler()` method. > > > > > > > > If you'll found a problem or something unexpected - write here or in > the > > > > ticket [1]. > > > > > > > > [1] https://issues.apache.org/jira/browse/IGNITE-8227 > > > > > > > >
