[
https://issues.apache.org/jira/browse/ACCUMULO-3100?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14126390#comment-14126390
]
Christopher Tubbs commented on ACCUMULO-3100:
---------------------------------------------
I mean, the problem affecting us was introduced in 2.6.0-SNAPSHOT.
The fact that it could affect users, if they fail to manage their dependencies
(patches, version) has existed for longer, yes, but there's little we can do
about that (except the kind of thing you did here).
The problem is that the Guava classes they're using should never have been used
(marked beta since their introduction and are inappropriate for a library like
Hadoop), are currently deprecated (in 14.0.1, with an alternative), and have
been showing up as warnings in their code for some time. These can be easily
patched to use non-beta, non-deprecated alternatives, like the Guava 14
javadocs instruct, without changing their dependency version. My question is if
anybody is already aware of a JIRA to do that (or to generally address the
thousands of Hadoop warnings which go overlooked and un-triaged, which this
would fall under).
It'd be nice for us, and for users in general, if they did use the
non-deprecated code in 14.0.1 so that the issue is mitigated for us, as well as
any users who need a newer version of Guava (because it's up to version 19
now). Guava does a good job of backwards-compatibility, so if this simple
action is taken, it should stabilize things for awhile, regardless of which
version we pick, or users need.
And, if a ticket already exists, this one should be linked against that one.
That's all I meant. I haven't found one, but didn't have time to look. I'll
look tomorrow if nobody else does, and create one if it doesn't exist, and link
this to it.
(sorry for the lack of brevity; trying to explain what I meant is hard)
> Accumulo fails to test against recent Hadoop 2.6.0-SNAPSHOT
> -----------------------------------------------------------
>
> Key: ACCUMULO-3100
> URL: https://issues.apache.org/jira/browse/ACCUMULO-3100
> Project: Accumulo
> Issue Type: Bug
> Components: test
> Affects Versions: 1.6.0
> Reporter: Josh Elser
> Assignee: Josh Elser
> Fix For: 1.6.1, 1.7.0
>
> Time Spent: 20m
> Remaining Estimate: 0h
>
> JobSubmitted makes a call out to CryptoUtils to test for encrypted shuffle
> support that was recently added to branch-2 (specifically HDFS-6134 and
> HADOOP-10150 looking at the blame)
> {noformat}
> java.lang.NoClassDefFoundError: com/google/common/io/LimitInputStream
> 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)
> at
> org.apache.hadoop.mapreduce.JobSubmitter.submitJobInternal(JobSubmitter.java:380)
> at org.apache.hadoop.mapreduce.Job$10.run(Job.java:1294)
> at org.apache.hadoop.mapreduce.Job$10.run(Job.java:1291)
> at java.security.AccessController.doPrivileged(Native Method)
> at javax.security.auth.Subject.doAs(Subject.java:415)
> at
> org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1614)
> at org.apache.hadoop.mapreduce.Job.submit(Job.java:1291)
> at org.apache.hadoop.mapreduce.Job.waitForCompletion(Job.java:1312)
> {noformat}
> Because of this, we can't run the test because we can't load LimitInputStream
> because we depend on Guava 15.0 which doesn't contain LimitInputStream.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
