本文共 3189 字,大约阅读时间需要 10 分钟。
线上一个ETL Job不能跑了,报异常,这里为了说明问题简化表结构:
1 2 3 4 | hive> desc void_t; OK x int None z void None |
而
1 | select * from void_t |
确实会抛出异常:
1 | 14 / 03 / 0201 : 28 : 58 ERROR CliDriver: Failed with exceptionjava.io.IOException:org.apache.hadoop.hive.ql.metadata.HiveException: Errorevaluating x |
看到这个异常很疑惑,和x字段有什么关系呢,继续看详细日志:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 | java.io.IOException:org.apache.hadoop.hive.ql.metadata.HiveException: Error evaluating x atorg.apache.hadoop.hive.ql.exec.FetchTask.fetch(FetchTask.java: 150 ) atorg.apache.hadoop.hive.ql.Driver.getResults(Driver.java: 1412 ) at org.apache.hadoop.hive.cli.CliDriver.processLocalCmd(CliDriver.java: 271 ) atorg.apache.hadoop.hive.cli.CliDriver.processCmd(CliDriver.java: 216 ) atorg.apache.hadoop.hive.cli.CliDriver.processLine(CliDriver.java: 413 ) at org.apache.hadoop.hive.cli.CliDriver.run(CliDriver.java: 756 ) atorg.apache.hadoop.hive.cli.CliDriver.main(CliDriver.java: 614 ) atsun.reflect.NativeMethodAccessorImpl.invoke0(Native Method) atsun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java: 39 ) atsun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java: 25 ) atjava.lang.reflect.Method.invoke(Method.java: 597 ) atorg.apache.hadoop.util.RunJar.main(RunJar.java: 208 ) Causedby: org.apache.hadoop.hive.ql.metadata.HiveException: Error evaluating x atorg.apache.hadoop.hive.ql.exec.SelectOperator.processOp(SelectOperator.java: 80 ) atorg.apache.hadoop.hive.ql.exec.Operator.process(Operator.java: 502 ) at org.apache.hadoop.hive.ql.exec.Operator.forward(Operator.java: 832 ) atorg.apache.hadoop.hive.ql.exec.TableScanOperator.processOp(TableScanOperator.java: 90 ) atorg.apache.hadoop.hive.ql.exec.Operator.process(Operator.java: 502 ) atorg.apache.hadoop.hive.ql.exec.FetchOperator.pushRow(FetchOperator.java: 490 ) atorg.apache.hadoop.hive.ql.exec.FetchTask.fetch(FetchTask.java: 136 ) ... 11 more Causedby: java.lang.RuntimeException: Internal error: no LazyObject for VOID atorg.apache.hadoop.hive.serde2.lazy.LazyFactory.createLazyPrimitiveClass(LazyFactory.java: 119 ) atorg.apache.hadoop.hive.serde2.lazy.LazyFactory.createLazyObject(LazyFactory.java: 155 ) at org.apache.hadoop.hive.serde2.lazy.LazyStruct.parse(LazyStruct.java: 108 ) atorg.apache.hadoop.hive.serde2.lazy.LazyStruct.getField(LazyStruct.java: 190 ) atorg.apache.hadoop.hive.serde2.lazy.objectinspector.LazySimpleStructObjectInspector.getStructFieldData(LazySimpleStructObjectInspector.java: 188 ) atorg.apache.hadoop.hive.serde2.objectinspector.DelegatedStructObjectInspector.getStructFieldData(DelegatedStructObjectInspector.java: 79 ) atorg.apache.hadoop.hive.ql.exec.ExprNodeColumnEvaluator.evaluate(ExprNodeColumnEvaluator.java: 98 ) atorg.apache.hadoop.hive.ql.exec.SelectOperator.processOp(SelectOperator.java: 76 ) |
看到这个noLazyObject for VOID才知道原来问题出现在这里,也就是字段z上;查看ETL Job里的Query发现里面一个建表的语句用到了create table xxx as select null as z from xxx这样的方式,进而生成了一个VOID类型的字段,但是Hive本身却无法处理该字段,在jira里确实也有这么一个unresolved的Bug:HIVE-2615
Workaround也比较简单:1.先建表再insert select 2.在ctas时cast(null as <type>) z来强制指定类型.
本文转自MIKE老毕 51CTO博客,原文链接:http://blog.51cto.com/boylook/1365747,如需转载请自行联系原作者