[jira] Commented: (PIG-733) Order by sampling dumps entire sample to hdfs which causes dfs FileSystem closed error on large input
[ https://issues.apache.org/jira/browse/PIG-733?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12696416#action_12696416 ] Giridharan Kesavan commented on PIG-733: I'm going to resubmit the patch to hudson .. Order by sampling dumps entire sample to hdfs which causes dfs FileSystem closed error on large input --- Key: PIG-733 URL: https://issues.apache.org/jira/browse/PIG-733 Project: Pig Issue Type: Bug Affects Versions: 0.2.0 Reporter: Pradeep Kamath Assignee: Pradeep Kamath Fix For: 0.3.0 Attachments: PIG-733-v2.patch, PIG-733.patch Order by has a sampling job which samples the input and creates a sorted list of sample items. CUrrently the number of items sampled is 100 per map task. So if the input is large resulting in many maps (say 50,000) the sample is big. This sorted sample is stored on dfs. The WeightedRangePartitioner computes quantile boundaries and weighted probabilities for repeating values in each map by reading the samples file from DFS. In queries with many maps (in the order of 50,000) the dfs read of the sample file fails with FileSystem closed error. This seems to point to a dfs issue wherein a big dfs file being read simultaneously by many dfs clients (in this case all maps) causes the clients to be closed. However on the pig side, loading the sample from each map in the final map reduce job and computing the quantile boundaries and weighted probabilities is inefficient. We should do this computation through a FindQuantiles udf in the same map reduce job which produces the sorted samples. This way lesser data is written to dfs and in the final map reduce job, the weightedRangePartitioner needs to just load the computed information. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-733) Order by sampling dumps entire sample to hdfs which causes dfs FileSystem closed error on large input
[ https://issues.apache.org/jira/browse/PIG-733?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12696442#action_12696442 ] Hadoop QA commented on PIG-733: --- -1 overall. Here are the results of testing the latest attachment http://issues.apache.org/jira/secure/attachment/12404769/PIG-733-v2.patch against trunk revision 759376. +1 @author. The patch does not contain any @author tags. -1 tests included. The patch doesn't appear to include any new or modified tests. Please justify why no tests are needed for this patch. +1 javadoc. The javadoc tool did not generate any warning messages. +1 javac. The applied patch does not increase the total number of javac compiler warnings. -1 findbugs. The patch appears to introduce 5 new Findbugs warnings. +1 release audit. The applied patch does not increase the total number of release audit warnings. -1 core tests. The patch failed core unit tests. +1 contrib tests. The patch passed contrib unit tests. Test results: http://hudson.zones.apache.org/hudson/job/Pig-Patch-minerva.apache.org/21/testReport/ Findbugs warnings: http://hudson.zones.apache.org/hudson/job/Pig-Patch-minerva.apache.org/21/artifact/trunk/build/test/findbugs/newPatchFindbugsWarnings.html Console output: http://hudson.zones.apache.org/hudson/job/Pig-Patch-minerva.apache.org/21/console This message is automatically generated. Order by sampling dumps entire sample to hdfs which causes dfs FileSystem closed error on large input --- Key: PIG-733 URL: https://issues.apache.org/jira/browse/PIG-733 Project: Pig Issue Type: Bug Affects Versions: 0.2.0 Reporter: Pradeep Kamath Assignee: Pradeep Kamath Fix For: 0.3.0 Attachments: PIG-733-v2.patch, PIG-733.patch Order by has a sampling job which samples the input and creates a sorted list of sample items. CUrrently the number of items sampled is 100 per map task. So if the input is large resulting in many maps (say 50,000) the sample is big. This sorted sample is stored on dfs. The WeightedRangePartitioner computes quantile boundaries and weighted probabilities for repeating values in each map by reading the samples file from DFS. In queries with many maps (in the order of 50,000) the dfs read of the sample file fails with FileSystem closed error. This seems to point to a dfs issue wherein a big dfs file being read simultaneously by many dfs clients (in this case all maps) causes the clients to be closed. However on the pig side, loading the sample from each map in the final map reduce job and computing the quantile boundaries and weighted probabilities is inefficient. We should do this computation through a FindQuantiles udf in the same map reduce job which produces the sorted samples. This way lesser data is written to dfs and in the final map reduce job, the weightedRangePartitioner needs to just load the computed information. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-733) Order by sampling dumps entire sample to hdfs which causes dfs FileSystem closed error on large input
[ https://issues.apache.org/jira/browse/PIG-733?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12696244#action_12696244 ] Pradeep Kamath commented on PIG-733: Tests are not included in this patch since there are existing tests for order by. All core unit tests did pass and finbugs gave the same number of warnings with and without the patch (output below). The excess warnings produced by the patch have been addressed in the new version of the patch (PIG-733-v2.patch). {noformat} === CORE UNIT TESTS OUTPUT WITH PATCH [prade...@afterside:/tmp/PIG-733/trunk] test-core: [mkdir] Created dir: /tmp/PIG-733/trunk/build/test/logs [junit] Running org.apache.pig.test.TestAdd [junit] Tests run: 1, Failures: 0, Errors: 0, Time elapsed: 0.056 sec ... [junit] Running org.apache.pig.test.TestTypeCheckingValidatorNoSchema [junit] Tests run: 13, Failures: 0, Errors: 0, Time elapsed: 0.629 sec [junit] Running org.apache.pig.test.TestUnion [junit] Tests run: 3, Failures: 0, Errors: 0, Time elapsed: 49.94 sec test-contrib: BUILD SUCCESSFUL Total time: 77 minutes 47 seconds === FINDBUGS OUTPUT WITH PATCH [prade...@afterside:/tmp/PIG-733/trunk] [prade...@chargesize:/tmp/PIG-733/trunk]ant -Dfindbugs.home=/homes/pradeepk/findbugs-1.3.8 findbugs Buildfile: build.xml ... findbugs: [mkdir] Created dir: /tmp/PIG-733/trunk/build/test/findbugs [findbugs] Executing findbugs from ant task [findbugs] Running FindBugs... [findbugs] Warnings generated: 665 [findbugs] Calculating exit code... [findbugs] Setting 'bugs found' flag (1) [findbugs] Exit code set to: 1 [findbugs] Java Result: 1 [findbugs] Output saved to /tmp/PIG-733/trunk/build/test/findbugs/pig-findbugs-report.xml [xslt] Processing /tmp/PIG-733/trunk/build/test/findbugs/pig-findbugs-report.xml to /tmp/PIG-733/trunk/build/test/findbugs/pig-findbugs-report.html [xslt] Loading stylesheet /homes/pradeepk/findbugs-1.3.8/src/xsl/default.xsl === FINDBUGS OUTPUT WITHOUT PATCH [prade...@chargesize:/tmp/svncheckout/trunk]ant -Dfindbugs.home=/homes/pradeepk/findbugs-1.3.8 findbugs Buildfile: build.xml check-for-findbugs: ... findbugs: [mkdir] Created dir: /tmp/svncheckout/trunk/build/test/findbugs [findbugs] Executing findbugs from ant task [findbugs] Running FindBugs... [findbugs] Warnings generated: 665 [findbugs] Calculating exit code... [findbugs] Setting 'bugs found' flag (1) [findbugs] Exit code set to: 1 [findbugs] Java Result: 1 [findbugs] Output saved to /tmp/svncheckout/trunk/build/test/findbugs/pig-findbugs-report.xml [xslt] Processing /tmp/svncheckout/trunk/build/test/findbugs/pig-findbugs-report.xml to /tmp/svncheckout/trunk/build/test/findbugs/pig-findbugs-report.html [xslt] Loading stylesheet /homes/pradeepk/findbugs-1.3.8/src/xsl/default.xsl {noformat} Order by sampling dumps entire sample to hdfs which causes dfs FileSystem closed error on large input --- Key: PIG-733 URL: https://issues.apache.org/jira/browse/PIG-733 Project: Pig Issue Type: Bug Affects Versions: 0.2.0 Reporter: Pradeep Kamath Assignee: Pradeep Kamath Fix For: 0.3.0 Attachments: PIG-733.patch Order by has a sampling job which samples the input and creates a sorted list of sample items. CUrrently the number of items sampled is 100 per map task. So if the input is large resulting in many maps (say 50,000) the sample is big. This sorted sample is stored on dfs. The WeightedRangePartitioner computes quantile boundaries and weighted probabilities for repeating values in each map by reading the samples file from DFS. In queries with many maps (in the order of 50,000) the dfs read of the sample file fails with FileSystem closed error. This seems to point to a dfs issue wherein a big dfs file being read simultaneously by many dfs clients (in this case all maps) causes the clients to be closed. However on the pig side, loading the sample from each map in the final map reduce job and computing the quantile boundaries and weighted probabilities is inefficient. We should do this computation through a FindQuantiles udf in the same map reduce job which produces the sorted samples. This way lesser data is written to dfs and in the final map reduce job, the weightedRangePartitioner needs to just load the computed information. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-733) Order by sampling dumps entire sample to hdfs which causes dfs FileSystem closed error on large input
[ https://issues.apache.org/jira/browse/PIG-733?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12694855#action_12694855 ] Hadoop QA commented on PIG-733: --- -1 overall. Here are the results of testing the latest attachment http://issues.apache.org/jira/secure/attachment/12404402/PIG-733.patch against trunk revision 759376. +1 @author. The patch does not contain any @author tags. -1 tests included. The patch doesn't appear to include any new or modified tests. Please justify why no tests are needed for this patch. +1 javadoc. The javadoc tool did not generate any warning messages. -1 javac. The applied patch generated 207 javac compiler warnings (more than the trunk's current 200 warnings). -1 findbugs. The patch appears to introduce 5 new Findbugs warnings. +1 release audit. The applied patch does not increase the total number of release audit warnings. -1 core tests. The patch failed core unit tests. +1 contrib tests. The patch passed contrib unit tests. Test results: http://hudson.zones.apache.org/hudson/job/Pig-Patch-minerva.apache.org/18/testReport/ Findbugs warnings: http://hudson.zones.apache.org/hudson/job/Pig-Patch-minerva.apache.org/18/artifact/trunk/build/test/findbugs/newPatchFindbugsWarnings.html Console output: http://hudson.zones.apache.org/hudson/job/Pig-Patch-minerva.apache.org/18/console This message is automatically generated. Order by sampling dumps entire sample to hdfs which causes dfs FileSystem closed error on large input --- Key: PIG-733 URL: https://issues.apache.org/jira/browse/PIG-733 Project: Pig Issue Type: Bug Affects Versions: 0.2.0 Reporter: Pradeep Kamath Assignee: Pradeep Kamath Fix For: 0.3.0 Attachments: PIG-733.patch Order by has a sampling job which samples the input and creates a sorted list of sample items. CUrrently the number of items sampled is 100 per map task. So if the input is large resulting in many maps (say 50,000) the sample is big. This sorted sample is stored on dfs. The WeightedRangePartitioner computes quantile boundaries and weighted probabilities for repeating values in each map by reading the samples file from DFS. In queries with many maps (in the order of 50,000) the dfs read of the sample file fails with FileSystem closed error. This seems to point to a dfs issue wherein a big dfs file being read simultaneously by many dfs clients (in this case all maps) causes the clients to be closed. However on the pig side, loading the sample from each map in the final map reduce job and computing the quantile boundaries and weighted probabilities is inefficient. We should do this computation through a FindQuantiles udf in the same map reduce job which produces the sorted samples. This way lesser data is written to dfs and in the final map reduce job, the weightedRangePartitioner needs to just load the computed information. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.