I reproduced the issue. The inconsistent test result is due to time zone dependency in the test.
The underlying issue is that convertToTimestamp attempts to set negative value with java.sql.Timestamp.setNanos https://issues.apache.org/jira/browse/FLINK-27465 Thanks, Thomas On Thu, Apr 28, 2022 at 7:23 PM Guowei Ma <guowei....@gmail.com> wrote: > Hi Haizhou > > I ran the test and there is no problem. > And commit is "d940af688be90c92ce4f8b9ca883f6753c94aa0f" > > Best, > Guowei > > > On Fri, Apr 29, 2022 at 5:39 AM Haizhou Zhao <zhaohaizhou940...@gmail.com> > wrote: > > > Hello Flink Community, > > > > I was encountering some unit test failure in the flink-avro sub-module > when > > I tried to pull down the master branch and build. > > > > Here is the command I ran: > > > > mvn clean package -pl flink-formats/flink-avro > > > > Here is the test that fails: > > > > > https://github.com/apache/flink/blob/master/flink-formats/flink-avro/src/test/java/org/apache/flink/formats/avro/AvroRowDeSerializationSchemaTest.java#L178 > > > > Here is the exception that was thrown: > > [ERROR] > > > > > org.apache.flink.formats.avro.AvroRowDeSerializationSchemaTest.testGenericDeserializeSeveralTimes > > Time elapsed: 0.008 s <<< ERROR! > > java.io.IOException: Failed to deserialize Avro record. > > ... > > > > Here is the latest commit of the HEAD I pulled: > > commit c5430e2e5d4eeb0aba14ce3ea8401747afe0182d (HEAD -> master, > > oss/master) > > > > Can someone confirm this is indeed a problem on the master branch? If > yes, > > any suggestions on fixing it? > > > > Thank you, > > Haizhou Zhao > > >