[jira] [Created] (MAPREDUCE-5604) TestMRAMWithNonNormalizedCapabilities fails on Windows due to exceeding max path length
Chris Nauroth created MAPREDUCE-5604: Summary: TestMRAMWithNonNormalizedCapabilities fails on Windows due to exceeding max path length Key: MAPREDUCE-5604 URL: https://issues.apache.org/jira/browse/MAPREDUCE-5604 Project: Hadoop Map/Reduce Issue Type: Bug Components: client, test Affects Versions: 2.2.0, 3.0.0 Reporter: Chris Nauroth Assignee: Chris Nauroth Priority: Minor The test uses the full class name as a component of the {{yarn.nodemanager.local-dirs}} setting for a {{MiniMRYarnCluster}}. This causes container launch to fail when trying to access files at a path longer than the maximum of 260 characters. -- This message was sent by Atlassian JIRA (v6.1#6144)
[jira] [Commented] (MAPREDUCE-5392) mapred job -history all command throws IndexOutOfBoundsException
[ https://issues.apache.org/jira/browse/MAPREDUCE-5392?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13809963#comment-13809963 ] Shinichi Yamashita commented on MAPREDUCE-5392: --- Jenkins occurred OutOfMemoryError. I checked Jenkins log and OOM occurred at native-maven-plugin phase. To make sure, I attach a same patch and Jenkins test. mapred job -history all command throws IndexOutOfBoundsException -- Key: MAPREDUCE-5392 URL: https://issues.apache.org/jira/browse/MAPREDUCE-5392 Project: Hadoop Map/Reduce Issue Type: Bug Components: mrv2 Affects Versions: 3.0.0, 2.0.5-alpha, 2.2.0 Reporter: Shinichi Yamashita Assignee: Shinichi Yamashita Priority: Minor Fix For: 3.0.0 Attachments: MAPREDUCE-5392.patch, MAPREDUCE-5392.patch, MAPREDUCE-5392.patch, MAPREDUCE-5392.patch, MAPREDUCE-5392.patch, MAPREDUCE-5392.patch, MAPREDUCE-5392.patch When I use an all option by mapred job -history comamnd, the following exceptions are displayed and do not work. {code} Exception in thread main java.lang.StringIndexOutOfBoundsException: String index out of range: -3 at java.lang.String.substring(String.java:1875) at org.apache.hadoop.mapreduce.util.HostUtil.convertTrackerNameToHostName(HostUtil.java:49) at org.apache.hadoop.mapreduce.jobhistory.HistoryViewer.getTaskLogsUrl(HistoryViewer.java:459) at org.apache.hadoop.mapreduce.jobhistory.HistoryViewer.printAllTaskAttempts(HistoryViewer.java:235) at org.apache.hadoop.mapreduce.jobhistory.HistoryViewer.print(HistoryViewer.java:117) at org.apache.hadoop.mapreduce.tools.CLI.viewHistory(CLI.java:472) at org.apache.hadoop.mapreduce.tools.CLI.run(CLI.java:313) at org.apache.hadoop.util.ToolRunner.run(ToolRunner.java:70) at org.apache.hadoop.util.ToolRunner.run(ToolRunner.java:84) at org.apache.hadoop.mapred.JobClient.main(JobClient.java:1233) {code} This is because a node name recorded in History file is not given tracker_. Therefore it makes modifications to be able to read History file even if a node name is not given by tracker_. In addition, it fixes the URL of displayed task log. -- This message was sent by Atlassian JIRA (v6.1#6144)
[jira] [Updated] (MAPREDUCE-5392) mapred job -history all command throws IndexOutOfBoundsException
[ https://issues.apache.org/jira/browse/MAPREDUCE-5392?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Shinichi Yamashita updated MAPREDUCE-5392: -- Attachment: MAPREDUCE-5392.patch mapred job -history all command throws IndexOutOfBoundsException -- Key: MAPREDUCE-5392 URL: https://issues.apache.org/jira/browse/MAPREDUCE-5392 Project: Hadoop Map/Reduce Issue Type: Bug Components: mrv2 Affects Versions: 3.0.0, 2.0.5-alpha, 2.2.0 Reporter: Shinichi Yamashita Assignee: Shinichi Yamashita Priority: Minor Fix For: 3.0.0 Attachments: MAPREDUCE-5392.patch, MAPREDUCE-5392.patch, MAPREDUCE-5392.patch, MAPREDUCE-5392.patch, MAPREDUCE-5392.patch, MAPREDUCE-5392.patch, MAPREDUCE-5392.patch, MAPREDUCE-5392.patch When I use an all option by mapred job -history comamnd, the following exceptions are displayed and do not work. {code} Exception in thread main java.lang.StringIndexOutOfBoundsException: String index out of range: -3 at java.lang.String.substring(String.java:1875) at org.apache.hadoop.mapreduce.util.HostUtil.convertTrackerNameToHostName(HostUtil.java:49) at org.apache.hadoop.mapreduce.jobhistory.HistoryViewer.getTaskLogsUrl(HistoryViewer.java:459) at org.apache.hadoop.mapreduce.jobhistory.HistoryViewer.printAllTaskAttempts(HistoryViewer.java:235) at org.apache.hadoop.mapreduce.jobhistory.HistoryViewer.print(HistoryViewer.java:117) at org.apache.hadoop.mapreduce.tools.CLI.viewHistory(CLI.java:472) at org.apache.hadoop.mapreduce.tools.CLI.run(CLI.java:313) at org.apache.hadoop.util.ToolRunner.run(ToolRunner.java:70) at org.apache.hadoop.util.ToolRunner.run(ToolRunner.java:84) at org.apache.hadoop.mapred.JobClient.main(JobClient.java:1233) {code} This is because a node name recorded in History file is not given tracker_. Therefore it makes modifications to be able to read History file even if a node name is not given by tracker_. In addition, it fixes the URL of displayed task log. -- This message was sent by Atlassian JIRA (v6.1#6144)
[jira] [Updated] (MAPREDUCE-5604) TestMRAMWithNonNormalizedCapabilities fails on Windows due to exceeding max path length
[ https://issues.apache.org/jira/browse/MAPREDUCE-5604?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Chris Nauroth updated MAPREDUCE-5604: - Attachment: MAPREDUCE-5604.1.patch I'm attaching a patch that applies the same fix we've used in similar cases: use the simple class name instead of the fullly qualified class name so that the testing directory is shorter. TestMRAMWithNonNormalizedCapabilities fails on Windows due to exceeding max path length --- Key: MAPREDUCE-5604 URL: https://issues.apache.org/jira/browse/MAPREDUCE-5604 Project: Hadoop Map/Reduce Issue Type: Bug Components: client, test Affects Versions: 3.0.0, 2.2.0 Reporter: Chris Nauroth Assignee: Chris Nauroth Priority: Minor Attachments: MAPREDUCE-5604.1.patch The test uses the full class name as a component of the {{yarn.nodemanager.local-dirs}} setting for a {{MiniMRYarnCluster}}. This causes container launch to fail when trying to access files at a path longer than the maximum of 260 characters. -- This message was sent by Atlassian JIRA (v6.1#6144)
[jira] [Updated] (MAPREDUCE-5604) TestMRAMWithNonNormalizedCapabilities fails on Windows due to exceeding max path length
[ https://issues.apache.org/jira/browse/MAPREDUCE-5604?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Chris Nauroth updated MAPREDUCE-5604: - Status: Patch Available (was: Open) TestMRAMWithNonNormalizedCapabilities fails on Windows due to exceeding max path length --- Key: MAPREDUCE-5604 URL: https://issues.apache.org/jira/browse/MAPREDUCE-5604 Project: Hadoop Map/Reduce Issue Type: Bug Components: client, test Affects Versions: 2.2.0, 3.0.0 Reporter: Chris Nauroth Assignee: Chris Nauroth Priority: Minor Attachments: MAPREDUCE-5604.1.patch The test uses the full class name as a component of the {{yarn.nodemanager.local-dirs}} setting for a {{MiniMRYarnCluster}}. This causes container launch to fail when trying to access files at a path longer than the maximum of 260 characters. -- This message was sent by Atlassian JIRA (v6.1#6144)
[jira] [Commented] (MAPREDUCE-5604) TestMRAMWithNonNormalizedCapabilities fails on Windows due to exceeding max path length
[ https://issues.apache.org/jira/browse/MAPREDUCE-5604?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13809977#comment-13809977 ] Hadoop QA commented on MAPREDUCE-5604: -- {color:red}-1 overall{color}. Here are the results of testing the latest attachment http://issues.apache.org/jira/secure/attachment/12611259/MAPREDUCE-5604.1.patch against trunk revision . {color:green}+1 @author{color}. The patch does not contain any @author tags. {color:green}+1 tests included{color}. The patch appears to include 1 new or modified test files. {color:red}-1 javac{color:red}. The patch appears to cause the build to fail. Console output: https://builds.apache.org/job/PreCommit-MAPREDUCE-Build/4160//console This message is automatically generated. TestMRAMWithNonNormalizedCapabilities fails on Windows due to exceeding max path length --- Key: MAPREDUCE-5604 URL: https://issues.apache.org/jira/browse/MAPREDUCE-5604 Project: Hadoop Map/Reduce Issue Type: Bug Components: client, test Affects Versions: 3.0.0, 2.2.0 Reporter: Chris Nauroth Assignee: Chris Nauroth Priority: Minor Attachments: MAPREDUCE-5604.1.patch The test uses the full class name as a component of the {{yarn.nodemanager.local-dirs}} setting for a {{MiniMRYarnCluster}}. This causes container launch to fail when trying to access files at a path longer than the maximum of 260 characters. -- This message was sent by Atlassian JIRA (v6.1#6144)
[jira] [Commented] (MAPREDUCE-5392) mapred job -history all command throws IndexOutOfBoundsException
[ https://issues.apache.org/jira/browse/MAPREDUCE-5392?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13809983#comment-13809983 ] Hadoop QA commented on MAPREDUCE-5392: -- {color:red}-1 overall{color}. Here are the results of testing the latest attachment http://issues.apache.org/jira/secure/attachment/12611258/MAPREDUCE-5392.patch against trunk revision . {color:green}+1 @author{color}. The patch does not contain any @author tags. {color:green}+1 tests included{color}. The patch appears to include 1 new or modified test files. {color:green}+1 javac{color}. The applied patch does not increase the total number of javac compiler warnings. {color:green}+1 javadoc{color}. The javadoc tool did not generate any warning messages. {color:green}+1 eclipse:eclipse{color}. The patch built with eclipse:eclipse. {color:green}+1 findbugs{color}. The patch does not introduce any new Findbugs (version 1.3.9) warnings. {color:green}+1 release audit{color}. The applied patch does not increase the total number of release audit warnings. {color:red}-1 core tests{color}. The patch failed these unit tests in hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-common hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-core hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-hs: org.apache.hadoop.mapreduce.v2.hs.TestJobHistoryParsing {color:green}+1 contrib tests{color}. The patch passed contrib unit tests. Test results: https://builds.apache.org/job/PreCommit-MAPREDUCE-Build/4161//testReport/ Console output: https://builds.apache.org/job/PreCommit-MAPREDUCE-Build/4161//console This message is automatically generated. mapred job -history all command throws IndexOutOfBoundsException -- Key: MAPREDUCE-5392 URL: https://issues.apache.org/jira/browse/MAPREDUCE-5392 Project: Hadoop Map/Reduce Issue Type: Bug Components: mrv2 Affects Versions: 3.0.0, 2.0.5-alpha, 2.2.0 Reporter: Shinichi Yamashita Assignee: Shinichi Yamashita Priority: Minor Fix For: 3.0.0 Attachments: MAPREDUCE-5392.patch, MAPREDUCE-5392.patch, MAPREDUCE-5392.patch, MAPREDUCE-5392.patch, MAPREDUCE-5392.patch, MAPREDUCE-5392.patch, MAPREDUCE-5392.patch, MAPREDUCE-5392.patch When I use an all option by mapred job -history comamnd, the following exceptions are displayed and do not work. {code} Exception in thread main java.lang.StringIndexOutOfBoundsException: String index out of range: -3 at java.lang.String.substring(String.java:1875) at org.apache.hadoop.mapreduce.util.HostUtil.convertTrackerNameToHostName(HostUtil.java:49) at org.apache.hadoop.mapreduce.jobhistory.HistoryViewer.getTaskLogsUrl(HistoryViewer.java:459) at org.apache.hadoop.mapreduce.jobhistory.HistoryViewer.printAllTaskAttempts(HistoryViewer.java:235) at org.apache.hadoop.mapreduce.jobhistory.HistoryViewer.print(HistoryViewer.java:117) at org.apache.hadoop.mapreduce.tools.CLI.viewHistory(CLI.java:472) at org.apache.hadoop.mapreduce.tools.CLI.run(CLI.java:313) at org.apache.hadoop.util.ToolRunner.run(ToolRunner.java:70) at org.apache.hadoop.util.ToolRunner.run(ToolRunner.java:84) at org.apache.hadoop.mapred.JobClient.main(JobClient.java:1233) {code} This is because a node name recorded in History file is not given tracker_. Therefore it makes modifications to be able to read History file even if a node name is not given by tracker_. In addition, it fixes the URL of displayed task log. -- This message was sent by Atlassian JIRA (v6.1#6144)
[jira] [Commented] (MAPREDUCE-5604) TestMRAMWithNonNormalizedCapabilities fails on Windows due to exceeding max path length
[ https://issues.apache.org/jira/browse/MAPREDUCE-5604?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13809987#comment-13809987 ] Chuan Liu commented on MAPREDUCE-5604: -- +1. Change looks good to me. TestMRAMWithNonNormalizedCapabilities fails on Windows due to exceeding max path length --- Key: MAPREDUCE-5604 URL: https://issues.apache.org/jira/browse/MAPREDUCE-5604 Project: Hadoop Map/Reduce Issue Type: Bug Components: client, test Affects Versions: 3.0.0, 2.2.0 Reporter: Chris Nauroth Assignee: Chris Nauroth Priority: Minor Attachments: MAPREDUCE-5604.1.patch The test uses the full class name as a component of the {{yarn.nodemanager.local-dirs}} setting for a {{MiniMRYarnCluster}}. This causes container launch to fail when trying to access files at a path longer than the maximum of 260 characters. -- This message was sent by Atlassian JIRA (v6.1#6144)
[jira] [Commented] (MAPREDUCE-5604) TestMRAMWithNonNormalizedCapabilities fails on Windows due to exceeding max path length
[ https://issues.apache.org/jira/browse/MAPREDUCE-5604?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13810414#comment-13810414 ] Chris Nauroth commented on MAPREDUCE-5604: -- bq. -1 javac. The patch appears to cause the build to fail. This is unrelated to the patch. It looks like another problem with the Jenkins server being overloaded: {code} Error occurred during initialization of VM Cannot create VM thread. Out of system resources. {code} TestMRAMWithNonNormalizedCapabilities fails on Windows due to exceeding max path length --- Key: MAPREDUCE-5604 URL: https://issues.apache.org/jira/browse/MAPREDUCE-5604 Project: Hadoop Map/Reduce Issue Type: Bug Components: client, test Affects Versions: 3.0.0, 2.2.0 Reporter: Chris Nauroth Assignee: Chris Nauroth Priority: Minor Attachments: MAPREDUCE-5604.1.patch The test uses the full class name as a component of the {{yarn.nodemanager.local-dirs}} setting for a {{MiniMRYarnCluster}}. This causes container launch to fail when trying to access files at a path longer than the maximum of 260 characters. -- This message was sent by Atlassian JIRA (v6.1#6144)
[jira] [Updated] (MAPREDUCE-5186) mapreduce.job.max.split.locations causes some splits created by CombineFileInputFormat to fail
[ https://issues.apache.org/jira/browse/MAPREDUCE-5186?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jason Lowe updated MAPREDUCE-5186: -- Target Version/s: 2.2.1 Status: Patch Available (was: Open) mapreduce.job.max.split.locations causes some splits created by CombineFileInputFormat to fail -- Key: MAPREDUCE-5186 URL: https://issues.apache.org/jira/browse/MAPREDUCE-5186 Project: Hadoop Map/Reduce Issue Type: Bug Components: job submission Affects Versions: 2.2.0, 2.0.4-alpha Reporter: Sangjin Lee Assignee: Robert Parker Priority: Critical Attachments: MAPREDUCE-5186v1.patch, MAPREDUCE-5186v2.patch, MAPREDUCE-5186v3.patch CombineFileInputFormat can easily create splits that can come from many different locations (during the last pass of creating global splits). However, we observe that this often runs afoul of the mapreduce.job.max.split.locations check that's done by JobSplitWriter. The default value for mapreduce.job.max.split.locations is 10, and with any decent size cluster, CombineFileInputFormat creates splits that are well above this limit. -- This message was sent by Atlassian JIRA (v6.1#6144)
[jira] [Updated] (MAPREDUCE-5186) mapreduce.job.max.split.locations causes some splits created by CombineFileInputFormat to fail
[ https://issues.apache.org/jira/browse/MAPREDUCE-5186?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jason Lowe updated MAPREDUCE-5186: -- Attachment: MAPREDUCE-5186v3.patch Updating Rob's patch with unit tests to verify truncation of locations is occurring when necessary. Also removed the TestBlockLimits test since it was checking for an exception in this case and is no longer necessary. mapreduce.job.max.split.locations causes some splits created by CombineFileInputFormat to fail -- Key: MAPREDUCE-5186 URL: https://issues.apache.org/jira/browse/MAPREDUCE-5186 Project: Hadoop Map/Reduce Issue Type: Bug Components: job submission Affects Versions: 2.0.4-alpha, 2.2.0 Reporter: Sangjin Lee Assignee: Robert Parker Priority: Critical Attachments: MAPREDUCE-5186v1.patch, MAPREDUCE-5186v2.patch, MAPREDUCE-5186v3.patch CombineFileInputFormat can easily create splits that can come from many different locations (during the last pass of creating global splits). However, we observe that this often runs afoul of the mapreduce.job.max.split.locations check that's done by JobSplitWriter. The default value for mapreduce.job.max.split.locations is 10, and with any decent size cluster, CombineFileInputFormat creates splits that are well above this limit. -- This message was sent by Atlassian JIRA (v6.1#6144)
[jira] [Commented] (MAPREDUCE-5186) mapreduce.job.max.split.locations causes some splits created by CombineFileInputFormat to fail
[ https://issues.apache.org/jira/browse/MAPREDUCE-5186?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13810631#comment-13810631 ] Sangjin Lee commented on MAPREDUCE-5186: Looks good to me. Thanks! mapreduce.job.max.split.locations causes some splits created by CombineFileInputFormat to fail -- Key: MAPREDUCE-5186 URL: https://issues.apache.org/jira/browse/MAPREDUCE-5186 Project: Hadoop Map/Reduce Issue Type: Bug Components: job submission Affects Versions: 2.0.4-alpha, 2.2.0 Reporter: Sangjin Lee Assignee: Robert Parker Priority: Critical Attachments: MAPREDUCE-5186v1.patch, MAPREDUCE-5186v2.patch, MAPREDUCE-5186v3.patch CombineFileInputFormat can easily create splits that can come from many different locations (during the last pass of creating global splits). However, we observe that this often runs afoul of the mapreduce.job.max.split.locations check that's done by JobSplitWriter. The default value for mapreduce.job.max.split.locations is 10, and with any decent size cluster, CombineFileInputFormat creates splits that are well above this limit. -- This message was sent by Atlassian JIRA (v6.1#6144)
[jira] [Commented] (MAPREDUCE-5186) mapreduce.job.max.split.locations causes some splits created by CombineFileInputFormat to fail
[ https://issues.apache.org/jira/browse/MAPREDUCE-5186?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13810639#comment-13810639 ] Hadoop QA commented on MAPREDUCE-5186: -- {color:red}-1 overall{color}. Here are the results of testing the latest attachment http://issues.apache.org/jira/secure/attachment/12611465/MAPREDUCE-5186v3.patch against trunk revision . {color:green}+1 @author{color}. The patch does not contain any @author tags. {color:green}+1 tests included{color}. The patch appears to include 1 new or modified test files. {color:red}-1 javac{color:red}. The patch appears to cause the build to fail. Console output: https://builds.apache.org/job/PreCommit-MAPREDUCE-Build/4162//console This message is automatically generated. mapreduce.job.max.split.locations causes some splits created by CombineFileInputFormat to fail -- Key: MAPREDUCE-5186 URL: https://issues.apache.org/jira/browse/MAPREDUCE-5186 Project: Hadoop Map/Reduce Issue Type: Bug Components: job submission Affects Versions: 2.0.4-alpha, 2.2.0 Reporter: Sangjin Lee Assignee: Robert Parker Priority: Critical Attachments: MAPREDUCE-5186v1.patch, MAPREDUCE-5186v2.patch, MAPREDUCE-5186v3.patch CombineFileInputFormat can easily create splits that can come from many different locations (during the last pass of creating global splits). However, we observe that this often runs afoul of the mapreduce.job.max.split.locations check that's done by JobSplitWriter. The default value for mapreduce.job.max.split.locations is 10, and with any decent size cluster, CombineFileInputFormat creates splits that are well above this limit. -- This message was sent by Atlassian JIRA (v6.1#6144)
[jira] [Updated] (MAPREDUCE-5186) mapreduce.job.max.split.locations causes some splits created by CombineFileInputFormat to fail
[ https://issues.apache.org/jira/browse/MAPREDUCE-5186?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jason Lowe updated MAPREDUCE-5186: -- Attachment: MAPREDUCE-5186v3.patch Apache build machine had a bunch of processes that had escaped and caused the Jenkins user to hit the process ulimit. Those have been cleaned up, so resubmitting the same patch to kick Jenkins again. mapreduce.job.max.split.locations causes some splits created by CombineFileInputFormat to fail -- Key: MAPREDUCE-5186 URL: https://issues.apache.org/jira/browse/MAPREDUCE-5186 Project: Hadoop Map/Reduce Issue Type: Bug Components: job submission Affects Versions: 2.0.4-alpha, 2.2.0 Reporter: Sangjin Lee Assignee: Robert Parker Priority: Critical Attachments: MAPREDUCE-5186v1.patch, MAPREDUCE-5186v2.patch, MAPREDUCE-5186v3.patch, MAPREDUCE-5186v3.patch CombineFileInputFormat can easily create splits that can come from many different locations (during the last pass of creating global splits). However, we observe that this often runs afoul of the mapreduce.job.max.split.locations check that's done by JobSplitWriter. The default value for mapreduce.job.max.split.locations is 10, and with any decent size cluster, CombineFileInputFormat creates splits that are well above this limit. -- This message was sent by Atlassian JIRA (v6.1#6144)
[jira] [Updated] (MAPREDUCE-5601) ShuffleHandler fadvises file regions as DONTNEED even when fetch fails
[ https://issues.apache.org/jira/browse/MAPREDUCE-5601?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Sandy Ryza updated MAPREDUCE-5601: -- Attachment: MAPREDUCE-5601.patch ShuffleHandler fadvises file regions as DONTNEED even when fetch fails -- Key: MAPREDUCE-5601 URL: https://issues.apache.org/jira/browse/MAPREDUCE-5601 Project: Hadoop Map/Reduce Issue Type: Improvement Affects Versions: 2.2.0 Reporter: Sandy Ryza Assignee: Sandy Ryza Attachments: MAPREDUCE-5601.patch, MAPREDUCE-5601.patch, MAPREDUCE-5601.patch When a reducer initiates a fetch request, it does not know whether it will be able to fit the fetched data in memory. The first part of the response tells how much data will be coming. If space is not currently available, the reduce will abandon its request and try again later. When this occurs, the ShuffleHandler still fadvises the file region as DONTNEED. Meaning that the next time it's asked for, it will definitely be read from disk, even if it happened to be in the page cache before the request. I noticed this when trying to figure out why my job was doing so much more disk IO in MR2 than in MR1. When I turned the fadvise stuff off, I found that disk reads went to nearly 0 on machines that had enough memory to fit map outputs into the page cache. I then straced the NodeManager and noticed that there were over four times as many fadvise DONTNEED calls as map-reduce pairs. Further logging showed the same map outputs being fetched about this many times. This is a regression from MR1, which only did the fadvise DONTNEED after all the bytes were transferred. -- This message was sent by Atlassian JIRA (v6.1#6144)
[jira] [Commented] (MAPREDUCE-5601) ShuffleHandler fadvises file regions as DONTNEED even when fetch fails
[ https://issues.apache.org/jira/browse/MAPREDUCE-5601?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13810809#comment-13810809 ] Hadoop QA commented on MAPREDUCE-5601: -- {color:red}-1 overall{color}. Here are the results of testing the latest attachment http://issues.apache.org/jira/secure/attachment/12611494/MAPREDUCE-5601.patch against trunk revision . {color:green}+1 @author{color}. The patch does not contain any @author tags. {color:red}-1 tests included{color}. The patch doesn't appear to include any new or modified tests. Please justify why no new tests are needed for this patch. Also please list what manual steps were performed to verify this patch. {color:green}+1 javac{color}. The applied patch does not increase the total number of javac compiler warnings. {color:green}+1 javadoc{color}. The javadoc tool did not generate any warning messages. {color:green}+1 eclipse:eclipse{color}. The patch built with eclipse:eclipse. {color:green}+1 findbugs{color}. The patch does not introduce any new Findbugs (version 1.3.9) warnings. {color:green}+1 release audit{color}. The applied patch does not increase the total number of release audit warnings. {color:green}+1 core tests{color}. The patch passed unit tests in hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-shuffle. {color:green}+1 contrib tests{color}. The patch passed contrib unit tests. Test results: https://builds.apache.org/job/PreCommit-MAPREDUCE-Build/4164//testReport/ Console output: https://builds.apache.org/job/PreCommit-MAPREDUCE-Build/4164//console This message is automatically generated. ShuffleHandler fadvises file regions as DONTNEED even when fetch fails -- Key: MAPREDUCE-5601 URL: https://issues.apache.org/jira/browse/MAPREDUCE-5601 Project: Hadoop Map/Reduce Issue Type: Improvement Affects Versions: 2.2.0 Reporter: Sandy Ryza Assignee: Sandy Ryza Attachments: MAPREDUCE-5601.patch, MAPREDUCE-5601.patch, MAPREDUCE-5601.patch When a reducer initiates a fetch request, it does not know whether it will be able to fit the fetched data in memory. The first part of the response tells how much data will be coming. If space is not currently available, the reduce will abandon its request and try again later. When this occurs, the ShuffleHandler still fadvises the file region as DONTNEED. Meaning that the next time it's asked for, it will definitely be read from disk, even if it happened to be in the page cache before the request. I noticed this when trying to figure out why my job was doing so much more disk IO in MR2 than in MR1. When I turned the fadvise stuff off, I found that disk reads went to nearly 0 on machines that had enough memory to fit map outputs into the page cache. I then straced the NodeManager and noticed that there were over four times as many fadvise DONTNEED calls as map-reduce pairs. Further logging showed the same map outputs being fetched about this many times. This is a regression from MR1, which only did the fadvise DONTNEED after all the bytes were transferred. -- This message was sent by Atlassian JIRA (v6.1#6144)
[jira] [Commented] (MAPREDUCE-5186) mapreduce.job.max.split.locations causes some splits created by CombineFileInputFormat to fail
[ https://issues.apache.org/jira/browse/MAPREDUCE-5186?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13810852#comment-13810852 ] Hadoop QA commented on MAPREDUCE-5186: -- {color:red}-1 overall{color}. Here are the results of testing the latest attachment http://issues.apache.org/jira/secure/attachment/12611483/MAPREDUCE-5186v3.patch against trunk revision . {color:green}+1 @author{color}. The patch does not contain any @author tags. {color:green}+1 tests included{color}. The patch appears to include 1 new or modified test files. {color:green}+1 javac{color}. The applied patch does not increase the total number of javac compiler warnings. {color:green}+1 javadoc{color}. The javadoc tool did not generate any warning messages. {color:green}+1 eclipse:eclipse{color}. The patch built with eclipse:eclipse. {color:green}+1 findbugs{color}. The patch does not introduce any new Findbugs (version 1.3.9) warnings. {color:green}+1 release audit{color}. The applied patch does not increase the total number of release audit warnings. {color:red}-1 core tests{color}. The following test timeouts occurred in hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-core hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-jobclient: org.apache.hadoop.mapreduce.v2.TestUberAM {color:green}+1 contrib tests{color}. The patch passed contrib unit tests. Test results: https://builds.apache.org/job/PreCommit-MAPREDUCE-Build/4163//testReport/ Console output: https://builds.apache.org/job/PreCommit-MAPREDUCE-Build/4163//console This message is automatically generated. mapreduce.job.max.split.locations causes some splits created by CombineFileInputFormat to fail -- Key: MAPREDUCE-5186 URL: https://issues.apache.org/jira/browse/MAPREDUCE-5186 Project: Hadoop Map/Reduce Issue Type: Bug Components: job submission Affects Versions: 2.0.4-alpha, 2.2.0 Reporter: Sangjin Lee Assignee: Robert Parker Priority: Critical Attachments: MAPREDUCE-5186v1.patch, MAPREDUCE-5186v2.patch, MAPREDUCE-5186v3.patch, MAPREDUCE-5186v3.patch CombineFileInputFormat can easily create splits that can come from many different locations (during the last pass of creating global splits). However, we observe that this often runs afoul of the mapreduce.job.max.split.locations check that's done by JobSplitWriter. The default value for mapreduce.job.max.split.locations is 10, and with any decent size cluster, CombineFileInputFormat creates splits that are well above this limit. -- This message was sent by Atlassian JIRA (v6.1#6144)
[jira] [Commented] (MAPREDUCE-5601) ShuffleHandler fadvises file regions as DONTNEED even when fetch fails
[ https://issues.apache.org/jira/browse/MAPREDUCE-5601?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13810873#comment-13810873 ] Todd Lipcon commented on MAPREDUCE-5601: bq. Or you're saying we would pass the amount of unreserved memory remaining? Yea... though could be problematic due to parallel fetches. I think best would be to add a new field to TaskAttemptCompletionEventProto which contains the size of the completed map output. Then the reducer scheduler could be smarter and avoid wasting the round trip on things which won't fit anyway. Given it's a PB, it could be done compatibly (and fall back to the current optimistic behavior). ShuffleHandler fadvises file regions as DONTNEED even when fetch fails -- Key: MAPREDUCE-5601 URL: https://issues.apache.org/jira/browse/MAPREDUCE-5601 Project: Hadoop Map/Reduce Issue Type: Improvement Affects Versions: 2.2.0 Reporter: Sandy Ryza Assignee: Sandy Ryza Attachments: MAPREDUCE-5601.patch, MAPREDUCE-5601.patch, MAPREDUCE-5601.patch When a reducer initiates a fetch request, it does not know whether it will be able to fit the fetched data in memory. The first part of the response tells how much data will be coming. If space is not currently available, the reduce will abandon its request and try again later. When this occurs, the ShuffleHandler still fadvises the file region as DONTNEED. Meaning that the next time it's asked for, it will definitely be read from disk, even if it happened to be in the page cache before the request. I noticed this when trying to figure out why my job was doing so much more disk IO in MR2 than in MR1. When I turned the fadvise stuff off, I found that disk reads went to nearly 0 on machines that had enough memory to fit map outputs into the page cache. I then straced the NodeManager and noticed that there were over four times as many fadvise DONTNEED calls as map-reduce pairs. Further logging showed the same map outputs being fetched about this many times. This is a regression from MR1, which only did the fadvise DONTNEED after all the bytes were transferred. -- This message was sent by Atlassian JIRA (v6.1#6144)
[jira] [Commented] (MAPREDUCE-5601) ShuffleHandler fadvises file regions as DONTNEED even when fetch fails
[ https://issues.apache.org/jira/browse/MAPREDUCE-5601?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13810875#comment-13810875 ] Todd Lipcon commented on MAPREDUCE-5601: Also, +1 for the patch ShuffleHandler fadvises file regions as DONTNEED even when fetch fails -- Key: MAPREDUCE-5601 URL: https://issues.apache.org/jira/browse/MAPREDUCE-5601 Project: Hadoop Map/Reduce Issue Type: Improvement Affects Versions: 2.2.0 Reporter: Sandy Ryza Assignee: Sandy Ryza Attachments: MAPREDUCE-5601.patch, MAPREDUCE-5601.patch, MAPREDUCE-5601.patch When a reducer initiates a fetch request, it does not know whether it will be able to fit the fetched data in memory. The first part of the response tells how much data will be coming. If space is not currently available, the reduce will abandon its request and try again later. When this occurs, the ShuffleHandler still fadvises the file region as DONTNEED. Meaning that the next time it's asked for, it will definitely be read from disk, even if it happened to be in the page cache before the request. I noticed this when trying to figure out why my job was doing so much more disk IO in MR2 than in MR1. When I turned the fadvise stuff off, I found that disk reads went to nearly 0 on machines that had enough memory to fit map outputs into the page cache. I then straced the NodeManager and noticed that there were over four times as many fadvise DONTNEED calls as map-reduce pairs. Further logging showed the same map outputs being fetched about this many times. This is a regression from MR1, which only did the fadvise DONTNEED after all the bytes were transferred. -- This message was sent by Atlassian JIRA (v6.1#6144)
[jira] [Commented] (MAPREDUCE-5601) ShuffleHandler fadvises file regions as DONTNEED even when fetch fails
[ https://issues.apache.org/jira/browse/MAPREDUCE-5601?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13810914#comment-13810914 ] Sandy Ryza commented on MAPREDUCE-5601: --- bq. Yea... though could be problematic due to parallel fetches. Right. It would help a little if the amount remaining was less than the total fetched in each request, but wouldn't solve the bigger problem. bq. I think best would be to add a new field to TaskAttemptCompletionEventProto which contains the size of the completed map output. That's been my thinking too. Unfortunately the task umbilical protocol is still on Writables, so there could be compatibility issues. ShuffleHandler fadvises file regions as DONTNEED even when fetch fails -- Key: MAPREDUCE-5601 URL: https://issues.apache.org/jira/browse/MAPREDUCE-5601 Project: Hadoop Map/Reduce Issue Type: Improvement Affects Versions: 2.2.0 Reporter: Sandy Ryza Assignee: Sandy Ryza Attachments: MAPREDUCE-5601.patch, MAPREDUCE-5601.patch, MAPREDUCE-5601.patch When a reducer initiates a fetch request, it does not know whether it will be able to fit the fetched data in memory. The first part of the response tells how much data will be coming. If space is not currently available, the reduce will abandon its request and try again later. When this occurs, the ShuffleHandler still fadvises the file region as DONTNEED. Meaning that the next time it's asked for, it will definitely be read from disk, even if it happened to be in the page cache before the request. I noticed this when trying to figure out why my job was doing so much more disk IO in MR2 than in MR1. When I turned the fadvise stuff off, I found that disk reads went to nearly 0 on machines that had enough memory to fit map outputs into the page cache. I then straced the NodeManager and noticed that there were over four times as many fadvise DONTNEED calls as map-reduce pairs. Further logging showed the same map outputs being fetched about this many times. This is a regression from MR1, which only did the fadvise DONTNEED after all the bytes were transferred. -- This message was sent by Atlassian JIRA (v6.1#6144)
[jira] [Updated] (MAPREDUCE-4362) If possible, we should get back the feature of propagating task logs back to JobClient
[ https://issues.apache.org/jira/browse/MAPREDUCE-4362?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Vinod Kumar Vavilapalli updated MAPREDUCE-4362: --- Target Version/s: 2.3.0 Sandy, can you rebase this patch? We need to remove the YARN changes as YARN-649 is in. May be some tests too? If possible, we should get back the feature of propagating task logs back to JobClient -- Key: MAPREDUCE-4362 URL: https://issues.apache.org/jira/browse/MAPREDUCE-4362 Project: Hadoop Map/Reduce Issue Type: Bug Components: client, mrv2 Affects Versions: 2.0.0-alpha Reporter: Vinod Kumar Vavilapalli Assignee: Sandy Ryza Attachments: MAPREDUCE-4362.patch, MAPREDUCE-4362.patch MAPREDUCE-3889 removed the code which was trying to pull from /tasklog. We should see if it is possible to get back the feature. -- This message was sent by Atlassian JIRA (v6.1#6144)