[HADOOP-9861] Invert ReflectionUtils' stack trace

• Type: Improvement
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 2.0.5-alpha
• Component/s: util
• Assignee: Daisuke Kobayashi
• Reporter: Harsh J
• Created: 10/Aug/13 07:54
• Updated: 12/Aug/13 16:04

Description

Often an MR task (as an example) may fail at the configure stage due to a misconfiguration or whatever, and the only thing a user gets by virtue of MR pulling limited bytes of the diagnostic error data is the top part of the stacktrace:

java.lang.RuntimeException: Error in configuring object
    at org.apache.hadoop.util.ReflectionUtils.setJobConf(ReflectionUtils.java:93)
    at org.apache.hadoop.util.ReflectionUtils.setConf(ReflectionUtils.java:64)
    at org.apache.hadoop.util.ReflectionUtils.newInstance(ReflectionUtils.java:117)
    at org.apache.hadoop.mapred.MapTask.runOldMapper(MapTask.java:432)
    at org.apache.hadoop.mapred.MapTask.run(MapTask.java:372)
    at org.apache.hadoop.mapred.Child$4.run(Child.java:255)

This is absolutely useless to a user, and he also goes ahead and blames the framework for having an issue, rather than thinking (non-intuitively) to go see the whole task log for the full trace, especially the last part.

Hundreds of time its been a mere class thats missing, etc. but there's just too much pain involved here to troubleshoot.

Would be much much better, if we inverted the trace. For example, here's what Hive can return back if we did so, for a random trouble I pulled from the web:

java.lang.RuntimeException: Error in configuring object
Caused by: java.lang.NullPointerException
    at org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector.toString(StructObjectInspector.java:64)
    at java.lang.String.valueOf(String.java:2826)
    at java.lang.StringBuilder.append(StringBuilder.java:115)
    at org.apache.hadoop.hive.ql.exec.UnionOperator.initializeOp(UnionOperator.java:110)
    at org.apache.hadoop.hive.ql.exec.Operator.initialize(Operator.java:375)
    at org.apache.hadoop.hive.ql.exec.Operator.initialize(Operator.java:451)
    at org.apache.hadoop.hive.ql.exec.Operator.initializeChildren(Operator.java:407)
    at org.apache.hadoop.hive.ql.exec.TableScanOperator.initializeOp(TableScanOperator.java:186)
    at org.apache.hadoop.hive.ql.exec.Operator.initialize(Operator.java:375)
    at org.apache.hadoop.hive.ql.exec.MapOperator.initializeOp(MapOperator.java:563)
    at org.apache.hadoop.hive.ql.exec.Operator.initialize(Operator.java:375)
    at org.apache.hadoop.hive.ql.exec.ExecMapper.configure(ExecMapper.java:100)
    ... 22 more

This way the user can at least be sure what part's really failing, and not get lost trying to work their way through reflection utils and upwards/downwards.