> On Aug. 30, 2018, 1:34 p.m., Peter Bacsko wrote: > > core/src/main/java/org/apache/oozie/action/hadoop/MapReduceActionExecutor.java > > Lines 575 (patched) > > <https://reviews.apache.org/r/68237/diff/6/?file=2077101#file2077101line575> > > > > Can't we just compare the IDs as strings? What's the rationale of this > > class? > > > > If it's not passed to other data structures or classes which use a > > Comparator<T>, then I guess it doesn't have to implement it.
The rationale behind this class is to compare two YARN application ID `String`s, when the cluster startup time is different. Newer application IDs will be treated as "greater". We cannot really use plain `String` comparison because we could face situations where different application IDs are read with different cluster startup times. - András ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/68237/#review208130 ----------------------------------------------------------- On Aug. 24, 2018, 2:18 p.m., András Piros wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/68237/ > ----------------------------------------------------------- > > (Updated Aug. 24, 2018, 2:18 p.m.) > > > Review request for oozie, Andras Salamon, Attila Sasvari, Kinga Marton, and > Peter Bacsko. > > > Repository: oozie-git > > > Description > ------- > > OOZIE-3298 OYA: external ID is not filled properly and failing MR job is > treated as SUCCEEDED > > > Diffs > ----- > > core/src/main/java/org/apache/oozie/action/hadoop/JavaActionExecutor.java > 8f0f244013932476d8ae454d224f235948529b34 > > core/src/main/java/org/apache/oozie/action/hadoop/MapReduceActionExecutor.java > 83a23f5220aa72ba15edc8b98ef80a74213fcee8 > > core/src/test/java/org/apache/oozie/action/hadoop/ActionExecutorTestCase.java > f39bba2c691435354dac6da7794e5142b511d937 > > core/src/test/java/org/apache/oozie/action/hadoop/TestJavaActionExecutor.java > a31079a41d30677d35a253a4a69505c21aa585f6 > > core/src/test/java/org/apache/oozie/action/hadoop/TestYarnApplicationIdComparator.java > PRE-CREATION > > core/src/test/java/org/apache/oozie/action/hadoop/TestYarnApplicationIdFinder.java > PRE-CREATION > > core/src/test/java/org/apache/oozie/action/oozie/TestSubWorkflowActionExecutor.java > 893405e38ad71d22460768b76ed03ac4d9a0b95d > > sharelib/streaming/src/test/java/org/apache/oozie/action/hadoop/TestMapReduceActionExecutor.java > f460b6bd11f60dfb397c6bba82be1427c2d1b570 > > > Diff: https://reviews.apache.org/r/68237/diff/6/ > > > Testing > ------- > > Tested on a real cluster, plus added test cases to > `TestMapReduceActionExecutor` and new test classes. > > Please note that I could find no proper way of actually getting MapReduce to > start a new job while using `OozieClient#submit()` - apparently no YARN child > application is created. Please advise what's the best way to advance, maybe > call `JobClient#submitJob()` directly from > `ActionExecutorTestCase#startWorkflowAndFailChildMRJob()`. > > > Thanks, > > András Piros > >
