|
|
@@ -54,11 +54,31 @@
|
|
|
<h3>Other Jiras (describe bug fixes and minor changes)</h3>
|
|
|
<ul>
|
|
|
|
|
|
+<li> <a href="https://issues.apache.org/jira/browse/HADOOP-8418">HADOOP-8418</a>.
|
|
|
+ Major bug reported by vicaya and fixed by crystal_gaoyu (security)<br>
|
|
|
+ <b>Fix UGI for IBM JDK running on Windows</b><br>
|
|
|
+ <blockquote>The login module and user principal classes are different for 32 and 64-bit Windows in IBM J9 JDK 6 SR10. Hadoop 1.0.3 does not run on either because it uses the 32 bit login module and the 64-bit user principal class.</blockquote></li>
|
|
|
+
|
|
|
+<li> <a href="https://issues.apache.org/jira/browse/HADOOP-8419">HADOOP-8419</a>.
|
|
|
+ Major bug reported by vicaya and fixed by carp84 (io)<br>
|
|
|
+ <b>GzipCodec NPE upon reset with IBM JDK</b><br>
|
|
|
+ <blockquote>The GzipCodec will NPE upon reset after finish when the native zlib codec is not loaded. When the native zlib is loaded the codec creates a CompressorOutputStream that doesn't have the problem, otherwise, the GZipCodec uses GZIPOutputStream which is extended to provide the resetState method. Since IBM JDK 6 SR9 FP2 including the current JDK 6 SR10, GZIPOutputStream#finish will release the underlying deflater, which causes NPE upon reset. This seems to be an IBM JDK quirk as Sun JDK and OpenJD...</blockquote></li>
|
|
|
+
|
|
|
+<li> <a href="https://issues.apache.org/jira/browse/HADOOP-8561">HADOOP-8561</a>.
|
|
|
+ Major improvement reported by vicaya and fixed by crystal_gaoyu (security)<br>
|
|
|
+ <b>Introduce HADOOP_PROXY_USER for secure impersonation in child hadoop client processes</b><br>
|
|
|
+ <blockquote>To solve the problem for an authenticated user to type hadoop shell commands in a web console, we can introduce an HADOOP_PROXY_USER environment variable to allow proper impersonation in the child hadoop client processes.</blockquote></li>
|
|
|
+
|
|
|
<li> <a href="https://issues.apache.org/jira/browse/HADOOP-8880">HADOOP-8880</a>.
|
|
|
Major bug reported by gkesavan and fixed by gkesavan <br>
|
|
|
<b>Missing jersey jars as dependency in the pom causes hive tests to fail</b><br>
|
|
|
<blockquote>ivy.xml has the dependency included where as the same dependency is not updated in the pom template.</blockquote></li>
|
|
|
|
|
|
+<li> <a href="https://issues.apache.org/jira/browse/HADOOP-9051">HADOOP-9051</a>.
|
|
|
+ Minor test reported by mgong@vmware.com and fixed by vicaya (test)<br>
|
|
|
+ <b>Òant testÓ will build failed for trying to delete a file</b><br>
|
|
|
+ <blockquote>Run "ant test" on branch-1 of hadoop-common.<br>When the test process reach "test-core-excluding-commit-and-smoke"<br><br>It will invoke the "macro-test-runner" to clear and rebuild the test environment.<br>Then the ant task command <delete dir="@{test.dir}/logs" /><br>failed for trying to delete an non-existent file.<br><br>following is the test result logs:<br>test-core-excluding-commit-and-smoke:<br> [delete] Deleting: /home/jdu/bdc/hadoop-topology-branch1-new/hadoop-common/build/test/testsfailed<br> [delete] Dele...</blockquote></li>
|
|
|
+
|
|
|
<li> <a href="https://issues.apache.org/jira/browse/HADOOP-9111">HADOOP-9111</a>.
|
|
|
Minor improvement reported by jingzhao and fixed by jingzhao (test)<br>
|
|
|
<b>Fix failed testcases with @ignore annotation In branch-1</b><br>
|
|
|
@@ -84,6 +104,21 @@
|
|
|
<b>"Text File Busy" errors launching MR tasks</b><br>
|
|
|
<blockquote>Some very small percentage of tasks fail with a "Text file busy" error.<br><br>The following was the original diagnosis:<br>{quote}<br>Our use of PrintWriter in TaskController.writeCommand is unsafe, since that class swallows all IO exceptions. We're not currently checking for errors, which I'm seeing result in occasional task failures with the message "Text file busy" - assumedly because the close() call is failing silently for some reason.<br>{quote}<br>.. but turned out to be another issue as well (see below)</blockquote></li>
|
|
|
|
|
|
+<li> <a href="https://issues.apache.org/jira/browse/MAPREDUCE-4272">MAPREDUCE-4272</a>.
|
|
|
+ Major bug reported by vicaya and fixed by crystal_gaoyu (task)<br>
|
|
|
+ <b>SortedRanges.Range#compareTo is not spec compliant</b><br>
|
|
|
+ <blockquote>SortedRanges.Range#compareTo does not satisfy the requirement of Comparable#compareTo, where "the implementor must ensure {noformat}sgn(x.compareTo(y)) == -sgn(y.compareTo(x)){noformat} for all x and y."<br><br>This is manifested as TestStreamingBadRecords failures in alternative JDKs.</blockquote></li>
|
|
|
+
|
|
|
+<li> <a href="https://issues.apache.org/jira/browse/MAPREDUCE-4396">MAPREDUCE-4396</a>.
|
|
|
+ Minor bug reported by vicaya and fixed by crystal_gaoyu (client)<br>
|
|
|
+ <b>Make LocalJobRunner work with private distributed cache</b><br>
|
|
|
+ <blockquote>Some LocalJobRunner related unit tests fails if user directory permission and/or umask is too restrictive.</blockquote></li>
|
|
|
+
|
|
|
+<li> <a href="https://issues.apache.org/jira/browse/MAPREDUCE-4397">MAPREDUCE-4397</a>.
|
|
|
+ Major improvement reported by vicaya and fixed by crystal_gaoyu (task-controller)<br>
|
|
|
+ <b>Introduce HADOOP_SECURITY_CONF_DIR for task-controller</b><br>
|
|
|
+ <blockquote>The linux task controller currently hard codes the directory in which to look for its config file at compile time (via the HADOOP_CONF_DIR macro). Adding a new environment variable to look for task-controller's conf dir (with strict permission checks) would make installation much more flexible.</blockquote></li>
|
|
|
+
|
|
|
<li> <a href="https://issues.apache.org/jira/browse/MAPREDUCE-4696">MAPREDUCE-4696</a>.
|
|
|
Minor bug reported by gopalv and fixed by gopalv <br>
|
|
|
<b>TestMRServerPorts throws NullReferenceException</b><br>
|
|
|
@@ -114,6 +149,10 @@
|
|
|
<b>TestRecoveryManager fails on branch-1</b><br>
|
|
|
<blockquote>Looks like the tests are extremely flaky and just hang.</blockquote></li>
|
|
|
|
|
|
+<li> <a href="https://issues.apache.org/jira/browse/MAPREDUCE-4888">MAPREDUCE-4888</a>.
|
|
|
+ Blocker bug reported by revans2 and fixed by vinodkv (mrv1)<br>
|
|
|
+ <b>NLineInputFormat drops data in 1.1 and beyond</b><br>
|
|
|
+ <blockquote>When trying to root cause why MAPREDUCE-4782 did not cause us issues on 1.0.2, I found out that HADOOP-7823 introduced essentially the exact same error into org.apache.hadoop.mapred.lib.NLineInputFormat.<br><br>In 1.X org.apache.hadoop.mapred.lib.NLineInputFormat and org.apache.hadoop.mapreduce.lib.input.NLineInputFormat are separate implementations. The latter had an off by one error in it until MAPREDUCE-4782 fixed it. The former had no error in it until HADOOP-7823 introduced it in 1.1 and MAPR...</blockquote></li>
|
|
|
|
|
|
</ul>
|
|
|
|
Matthew Foley