[SPARK-11125] [SQL] Uninformative exception when running spark-sql witho…

[zjffdu]

…ut building with -Phive-thriftserver and SPARK_PREPEND_CLASSES is set

This is the exception after this patch. Please help review.

java.lang.NoClassDefFoundError: org/apache/hadoop/hive/cli/CliDriver
    at java.lang.ClassLoader.defineClass1(Native Method)
    at java.lang.ClassLoader.defineClass(ClassLoader.java:800)
    at java.security.SecureClassLoader.defineClass(SecureClassLoader.java:142)
    at java.net.URLClassLoader.defineClass(URLClassLoader.java:449)
    at java.net.URLClassLoader.access$100(URLClassLoader.java:71)
    at java.net.URLClassLoader$1.run(URLClassLoader.java:361)
    at java.net.URLClassLoader$1.run(URLClassLoader.java:355)
    at java.security.AccessController.doPrivileged(Native Method)
    at java.net.URLClassLoader.findClass(URLClassLoader.java:354)
    at java.lang.ClassLoader.loadClass(ClassLoader.java:425)
    at sun.misc.Launcher$AppClassLoader.loadClass(Launcher.java:308)
    at java.lang.ClassLoader.loadClass(ClassLoader.java:412)
    at java.lang.ClassLoader.loadClass(ClassLoader.java:358)
    at java.lang.Class.forName0(Native Method)
    at java.lang.Class.forName(Class.java:270)
    at org.apache.spark.util.Utils$.classForName(Utils.scala:173)
    at org.apache.spark.deploy.SparkSubmit$.org$apache$spark$deploy$SparkSubmit$$runMain(SparkSubmit.scala:647)
    at org.apache.spark.deploy.SparkSubmit$.doRunMain$1(SparkSubmit.scala:180)
    at org.apache.spark.deploy.SparkSubmit$.submit(SparkSubmit.scala:205)
    at org.apache.spark.deploy.SparkSubmit$.main(SparkSubmit.scala:120)
    at org.apache.spark.deploy.SparkSubmit.main(SparkSubmit.scala)
Caused by: java.lang.ClassNotFoundException: org.apache.hadoop.hive.cli.CliDriver
    at java.net.URLClassLoader$1.run(URLClassLoader.java:366)
    at java.net.URLClassLoader$1.run(URLClassLoader.java:355)
    at java.security.AccessController.doPrivileged(Native Method)
    at java.net.URLClassLoader.findClass(URLClassLoader.java:354)
    at java.lang.ClassLoader.loadClass(ClassLoader.java:425)
    at sun.misc.Launcher$AppClassLoader.loadClass(Launcher.java:308)
    at java.lang.ClassLoader.loadClass(ClassLoader.java:358)
    ... 21 more
Failed to load hive class.
You need to build Spark with -Phive and -Phive-thriftserver.

[SparkQA]

Test build #43774 has finished for PR 9134 at commit 2bdf979.

• This patch fails Scala style tests.
• This patch merges cleanly.
• This patch adds no public classes.

[SparkQA]

Test build #43783 has finished for PR 9134 at commit 5b6e651.

• This patch fails Spark unit tests.
• This patch merges cleanly.
• This patch adds no public classes.

[zjffdu]

Check the test report. Seems the failed unit tests are unrelated.

[srowen]

I think this OK. (PS the exception isn't "unreadable" -- just "uninformative"). NoClassDefFoundError is lower-level problem; classes were compiled in the presence of some other classes that aren't available at runtime. It's "worse" than ClassNotFoundException. It probably makes sense, but is this really a problem of building without Hive support? I would kind of think it's the opposite: you built with Hive and then it's not available at runtime?

[andrewor14]

Are we certain that we only get NoClassDefFoundError but not also ClassNotFoundException? I wonder if we should just check for either of those in the match. Also cc @yhuai

[srowen]

@zjffdu given your similar PR on this topic, can you address the question of how this occurs? The "Error" is a different lower-level thing.

[zjffdu]

@andrewor14 @srowen Here's how to reproduce the issue.

1. export SPARK_PREPEND_CLASSES=true
2. Build spark with command: build/sbt assembly -Pyarn -Phadoop-2.6 -Phive -Phive-thriftserver
3. Build spark again with command: build/sbt assembly -Pyarn -Phadoop-2.6
4. Start spark-sql: bin/spark-sql

The root cause is that we still have SparkSQLCLIDriver in classpath (under sql/hive-thriftserver/target). But we don't have hive related classes in classpath.

[srowen]

Hm, it still sounds like it's more of an artifact of mixing build artifacts. Something is being loaded that was compiled with Hive support, but then finds it's actually missing. Even though that's an error you'd only encounter as a developer (?) I suppose that's still a good motive. OK by me.

[zjffdu]

Yes, that would be encountered as developer

[yhuai]

ok to test

[yhuai]

test this please

[SparkQA]

Test build #44232 has finished for PR 9134 at commit 5b6e651.

• This patch fails Spark unit tests.
• This patch merges cleanly.
• This patch adds no public classes.

[zjffdu]

Seems the failed test is not related

[yhuai]

test this please

[SparkQA]

Test build #44286 has finished for PR 9134 at commit 5b6e651.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[yhuai]

Thanks! Merging to master.