[ 
https://issues.apache.org/jira/browse/HADOOP-12611?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15042024#comment-15042024
 ] 

Robert Kanter commented on HADOOP-12611:
----------------------------------------

{quote} it can inspect the initial state of secret (currentSecret\[1] == null) 
right after the three ZKSignerSecretProvider are initialized.{quote}
Ok, I think the problem here is that HADOOP-12181 changed the rollover 
frequency to only 2 seconds (it was originally 15).  So, it's very likely that 
one or more of the {{ZKSignerSecretProviders}} has already rolled by the time 
we check that {{currentSecret\[1] == null}}, so it's no longer {{null}}.  With 
a rollover time of 15 seconds, this was much less likely to occur, though I 
suppose it was still possible.

Trying to keep the timing/randomness that we'd like to test but still making 
the test not flakey, I think the best approach would be to do something similar 
to the 001 patch where you'd have a subclass of {{ZKSignerSecretProvider}} that 
keeps track of {{secrets}}, let it run for two iterations (which you can check 
based on the size of the list keeping the {{secrets}} so no timing is 
involved), and then verify that each iteration of {{secrets}} is correct.  For 
example:
# {{\[s1]\[null]}}
# {{\[s2]\[s1]}}
# {{\[s3]\[s2]}}

And the test itself doesn't have any timing components because you'd be waiting 
for the list of {{secrets}} from each {{ZKSignerSecretProvider}} to contain 3 
snapshots.  You can then just check that all sets of snapshots agree.

Does that sound good?

> TestZKSignerSecretProvider#testMultipleInit occasionally fail
> -------------------------------------------------------------
>
>                 Key: HADOOP-12611
>                 URL: https://issues.apache.org/jira/browse/HADOOP-12611
>             Project: Hadoop Common
>          Issue Type: Bug
>            Reporter: Wei-Chiu Chuang
>            Assignee: Wei-Chiu Chuang
>         Attachments: HADOOP-12611.001.patch, HADOOP-12611.002.patch
>
>
> https://builds.apache.org/job/Hadoop-Common-trunk/2053/testReport/junit/org.apache.hadoop.security.authentication.util/TestZKSignerSecretProvider/testMultipleInit/
> Error Message
> expected null, but was:<[B@142bad79>
> Stacktrace
> java.lang.AssertionError: expected null, but was:<[B@142bad79>
>       at org.junit.Assert.fail(Assert.java:88)
>       at org.junit.Assert.failNotNull(Assert.java:664)
>       at org.junit.Assert.assertNull(Assert.java:646)
>       at org.junit.Assert.assertNull(Assert.java:656)
>       at 
> org.apache.hadoop.security.authentication.util.TestZKSignerSecretProvider.testMultipleInit(TestZKSignerSecretProvider.java:149)
> I think the failure was introduced after HADOOP-12181
> This is likely where the root cause is:
> 2015-11-29 00:24:33,325 ERROR ZKSignerSecretProvider - An unexpected 
> exception occurred while pulling data fromZooKeeper
> java.lang.IllegalStateException: instance must be started before calling this 
> method
>       at 
> com.google.common.base.Preconditions.checkState(Preconditions.java:145)
>       at 
> org.apache.curator.framework.imps.CuratorFrameworkImpl.getData(CuratorFrameworkImpl.java:363)
>       at 
> org.apache.hadoop.security.authentication.util.ZKSignerSecretProvider.pullFromZK(ZKSignerSecretProvider.java:341)
>       at 
> org.apache.hadoop.security.authentication.util.ZKSignerSecretProvider.rollSecret(ZKSignerSecretProvider.java:264)
>       at 
> org.apache.hadoop.security.authentication.util.ZKSignerSecretProvider$$EnhancerByMockitoWithCGLIB$$575f06d8.CGLIB$rollSecret$2(<generated>)
>       at 
> org.apache.hadoop.security.authentication.util.ZKSignerSecretProvider$$EnhancerByMockitoWithCGLIB$$575f06d8$$FastClassByMockitoWithCGLIB$$6f94a716.invoke(<generated>)
>       at org.mockito.cglib.proxy.MethodProxy.invokeSuper(MethodProxy.java:216)
>       at 
> org.mockito.internal.creation.AbstractMockitoMethodProxy.invokeSuper(AbstractMockitoMethodProxy.java:10)
>       at 
> org.mockito.internal.invocation.realmethod.CGLIBProxyRealMethod.invoke(CGLIBProxyRealMethod.java:22)
>       at 
> org.mockito.internal.invocation.realmethod.FilteredCGLIBProxyRealMethod.invoke(FilteredCGLIBProxyRealMethod.java:27)
>       at 
> org.mockito.internal.invocation.Invocation.callRealMethod(Invocation.java:211)
>       at 
> org.mockito.internal.stubbing.answers.CallsRealMethods.answer(CallsRealMethods.java:36)
>       at org.mockito.internal.MockHandler.handle(MockHandler.java:99)
>       at 
> org.mockito.internal.creation.MethodInterceptorFilter.intercept(MethodInterceptorFilter.java:47)
>       at 
> org.apache.hadoop.security.authentication.util.ZKSignerSecretProvider$$EnhancerByMockitoWithCGLIB$$575f06d8.rollSecret(<generated>)
>       at 
> org.apache.hadoop.security.authentication.util.RolloverSignerSecretProvider$1.run(RolloverSignerSecretProvider.java:97)
>       at 
> java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:471)
>       at java.util.concurrent.FutureTask.runAndReset(FutureTask.java:304)
>       at 
> java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$301(ScheduledThreadPoolExecutor.java:178)
>       at 
> java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:293)
>       at 
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
>       at 
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)
>       at java.lang.Thread.run(Thread.java:745)



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to