[jira] Updated: (HDFS-311) Modifications to enable multiple types of logging
[ https://issues.apache.org/jira/browse/HDFS-311?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Luca Telloli updated HDFS-311: -- Attachment: HDFS-311-complete.patch Modifications to enable multiple types of logging -- Key: HDFS-311 URL: https://issues.apache.org/jira/browse/HDFS-311 Project: Hadoop HDFS Issue Type: Improvement Reporter: Luca Telloli Assignee: Luca Telloli Attachments: HADOOP-5188.patch, HADOOP-5188.patch, HADOOP-5188.patch, HADOOP-5188.patch, HADOOP-5188.patch, HADOOP-5188.patch, HADOOP-5188.pdf, HDFS-311-complete.patch, HDFS-311.patch -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (HDFS-311) Modifications to enable multiple types of logging
[ https://issues.apache.org/jira/browse/HDFS-311?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Luca Telloli updated HDFS-311: -- Attachment: HDFS-311-complete.patch Modifications to enable multiple types of logging -- Key: HDFS-311 URL: https://issues.apache.org/jira/browse/HDFS-311 Project: Hadoop HDFS Issue Type: Improvement Reporter: Luca Telloli Assignee: Luca Telloli Attachments: HADOOP-5188.patch, HADOOP-5188.patch, HADOOP-5188.patch, HADOOP-5188.patch, HADOOP-5188.patch, HADOOP-5188.patch, HADOOP-5188.pdf, HDFS-311-complete.patch, HDFS-311.patch, HDFS-311.patch -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (HDFS-234) Integration with BookKeeper logging system
[ https://issues.apache.org/jira/browse/HDFS-234?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Luca Telloli updated HDFS-234: -- Attachment: HDFS-234.patch Integration with BookKeeper logging system -- Key: HDFS-234 URL: https://issues.apache.org/jira/browse/HDFS-234 Project: Hadoop HDFS Issue Type: New Feature Reporter: Luca Telloli Assignee: Luca Telloli Attachments: create.png, HADOOP-5189-trunk-preview.patch, HADOOP-5189-trunk-preview.patch, HADOOP-5189-trunk-preview.patch, HADOOP-5189-v.19.patch, HADOOP-5189.patch, HDFS-234.patch BookKeeper is a system to reliably log streams of records (https://issues.apache.org/jira/browse/ZOOKEEPER-276). The NameNode is a natural target for such a system for being the metadata repository of the entire file system for HDFS. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (HDFS-487) HDFS should expose a fileid to uniquely identify a file
[ https://issues.apache.org/jira/browse/HDFS-487?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12730447#action_12730447 ] Hong Tang commented on HDFS-487: How about using [UUID|http://en.wikipedia.org/wiki/Universally_Unique_Identifier]? They have advantages: - can be calculated independently by any process. - guaranteed to be unique globally, even across two different HDFS instances, which would make it easier if we want to build a federated file system. HDFS should expose a fileid to uniquely identify a file --- Key: HDFS-487 URL: https://issues.apache.org/jira/browse/HDFS-487 Project: Hadoop HDFS Issue Type: New Feature Reporter: dhruba borthakur Assignee: dhruba borthakur HDFS should expose a id that uniquely identifies a file. This helps in developing applications that work correctly even when files are moved from one directory to another. A typical use-case is to make the Pluggable Block Placement Policy (HDFS-385) use fileid instead of filename. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Created: (HDFS-489) Updated TestHDFSCLI for changes from HADOOP-6139
Updated TestHDFSCLI for changes from HADOOP-6139 Key: HDFS-489 URL: https://issues.apache.org/jira/browse/HDFS-489 Project: Hadoop HDFS Issue Type: Bug Reporter: Jakob Homan Assignee: Jakob Homan HADOOP-6139 changed the output of the rm/rmr console text. The unit test for this is in hdfs, so TestHDFSCLI needs to be updated for the new output. Patch shortly. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (HDFS-397) Incorporate storage directories into EditLogFileInput/Output streams
[ https://issues.apache.org/jira/browse/HDFS-397?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12730464#action_12730464 ] Raghu Angadi commented on HDFS-397: --- Regd 1) : What about 'STORAGE_JSPOOL_DIR' which points to a different directory from 'current'. JSPOOL_FILE also 'happens' to be same but intention of original author might have been to be able change. We should remove these constants if these are not sued. bq. I assumed it's correct since it passed the unit tests correctly. hmm.. I don't think that is sufficient or wise. When I program I would like to know what I am doing and why it is correct at every line. Have a look for instance to line 1497 and following. Is it for the latest patch attached there? I am asking this mainly so that I don't have go through the big patch for HDFS-311. Incorporate storage directories into EditLogFileInput/Output streams Key: HDFS-397 URL: https://issues.apache.org/jira/browse/HDFS-397 Project: Hadoop HDFS Issue Type: Sub-task Reporter: Luca Telloli Assignee: Luca Telloli Attachments: HADOOP-6001.patch, HADOOP-6001.patch, HADOOP-6001.patch, HDFS-397.patch -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (HDFS-234) Integration with BookKeeper logging system
[ https://issues.apache.org/jira/browse/HDFS-234?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12730481#action_12730481 ] Flavio Paiva Junqueira commented on HDFS-234: - When you read from a BookKeeper ledger, it seems that you are reading the whole ledger before processing it. I can see two potential problems with it: # If the ledger is large, then the process reading the ledger might end up consuming too much memory; # It does not overlap reading the ledger with processing. It might be best to try to overlap reading from the ledger with processing the edits. I don't know what changes this suggestion implies, though. Integration with BookKeeper logging system -- Key: HDFS-234 URL: https://issues.apache.org/jira/browse/HDFS-234 Project: Hadoop HDFS Issue Type: New Feature Reporter: Luca Telloli Assignee: Luca Telloli Attachments: create.png, HADOOP-5189-trunk-preview.patch, HADOOP-5189-trunk-preview.patch, HADOOP-5189-trunk-preview.patch, HADOOP-5189-v.19.patch, HADOOP-5189.patch, HDFS-234.patch, zookeeper-dev-bookkeeper.jar, zookeeper-dev.jar BookKeeper is a system to reliably log streams of records (https://issues.apache.org/jira/browse/ZOOKEEPER-276). The NameNode is a natural target for such a system for being the metadata repository of the entire file system for HDFS. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (HDFS-462) Unit tests not working under Windows
[ https://issues.apache.org/jira/browse/HDFS-462?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12730518#action_12730518 ] Jakob Homan commented on HDFS-462: -- Again, no new tests as is fixing a defect that's being caught by the current unit tests (under Windows at least). Patch is ready to go. Unit tests not working under Windows Key: HDFS-462 URL: https://issues.apache.org/jira/browse/HDFS-462 Project: Hadoop HDFS Issue Type: Bug Components: name-node, test Affects Versions: 0.21.0 Environment: Windows Reporter: Luca Telloli Assignee: Jakob Homan Fix For: 0.21.0 Attachments: HDFS-462.patch, TEST-org.apache.hadoop.hdfs.TestHdfs.txt Unit tests are failing on windows due to a problem with rename. The failing code is around line 520 in FSImage.java: {noformat} assert curDir.exists() : Current directory must exist.; assert !prevDir.exists() : prvious directory must not exist.; assert !tmpDir.exists() : prvious.tmp directory must not exist.; // rename current to tmp rename(curDir, tmpDir); // save new image if (!curDir.mkdir()) throw new IOException(Cannot create directory + curDir); {noformat} and seems related to some open file or directory -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (HDFS-458) Create target for 10 minute patch test build for hdfs
[ https://issues.apache.org/jira/browse/HDFS-458?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jakob Homan updated HDFS-458: - Status: Open (was: Patch Available) Canceling patch to address Nicholas' comments. Create target for 10 minute patch test build for hdfs - Key: HDFS-458 URL: https://issues.apache.org/jira/browse/HDFS-458 Project: Hadoop HDFS Issue Type: New Feature Components: test Reporter: Jakob Homan Assignee: Jakob Homan Attachments: build.xml, HDFS-458.patch, TenMinuteTestData.xlsx It would be good to identify a subset of hdfs tests that provide strong test code coverage within 10 minutes, as is the goal of MAPREDUCE-670 and HADOOP-5628. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (HDFS-489) Updated TestHDFSCLI for changes from HADOOP-6139
[ https://issues.apache.org/jira/browse/HDFS-489?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jakob Homan updated HDFS-489: - Status: Patch Available (was: Open) submitting patch Updated TestHDFSCLI for changes from HADOOP-6139 Key: HDFS-489 URL: https://issues.apache.org/jira/browse/HDFS-489 Project: Hadoop HDFS Issue Type: Bug Reporter: Jakob Homan Assignee: Jakob Homan Attachments: HDFS-489.patch HADOOP-6139 changed the output of the rm/rmr console text. The unit test for this is in hdfs, so TestHDFSCLI needs to be updated for the new output. Patch shortly. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (HDFS-489) Updated TestHDFSCLI for changes from HADOOP-6139
[ https://issues.apache.org/jira/browse/HDFS-489?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jakob Homan updated HDFS-489: - Attachment: HDFS-489.patch Patch fixes TestHDFSCLI with new expected behavior. Even though the original patch was backported to 20, this particular test case was added after 20, so there is no need for a 20 port of this patch. It looks like the commons jar has been updated in the git repo, so TestHDFSCLI is currently breaking with this behavior and should be fixed. Hudson will complain about no tests but this does modify a test. Updated TestHDFSCLI for changes from HADOOP-6139 Key: HDFS-489 URL: https://issues.apache.org/jira/browse/HDFS-489 Project: Hadoop HDFS Issue Type: Bug Reporter: Jakob Homan Assignee: Jakob Homan Attachments: HDFS-489.patch HADOOP-6139 changed the output of the rm/rmr console text. The unit test for this is in hdfs, so TestHDFSCLI needs to be updated for the new output. Patch shortly. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (HDFS-475) Create a separate targets for fault injection related test and jar files creation files
[ https://issues.apache.org/jira/browse/HDFS-475?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Tsz Wo (Nicholas), SZE updated HDFS-475: Hadoop Flags: (was: [Reviewed]) In HDFS-483, I implemented some tests which use fi. I have to change the test target as following in order to run the tests. Do you want to add them to your patch? {code} @@ -318,7 +318,7 @@ !-- Weaving aspects in place Later on one can run 'ant jar' to create Hadoop jar file with instrumented classes -- - target name=injectfaults depends=compile description=Weaves aspects into precomplied HDFS classes + target name=injectfaults depends=compile, compile-hdfs-test description=Weaves aspects into precomplied HDFS classes !-- AspectJ task definition -- taskdef resource=org/aspectj/tools/ant/taskdefs/aspectjTaskdefs.properties classpath @@ -335,7 +335,7 @@ target=${javac.version} source=${javac.version} deprecation=${javac.deprecation} -classpath refid=classpath / +classpath refid=test.classpath / /iajc echo message=Weaving of aspects is finished/ /target @@ -500,10 +500,14 @@ batchtest todir=${test.build.dir} unless=testcase fileset dir=${test.src.dir}/hdfs includes=**/${test.include}.java - excludes=**/${test.exclude}.java / + excludes=**/${test.exclude}.java / +fileset dir=${test.src.dir}/aop + includes=**/${test.include}.java + excludes=**/${test.exclude}.java / /batchtest batchtest todir=${test.build.dir} if=testcase fileset dir=${test.src.dir}/hdfs includes=**/${testcase}.java/ +fileset dir=${test.src.dir}/aop includes=**/${testcase}.java/ /batchtest /junit antcall target=checkfailure/ {code} Create a separate targets for fault injection related test and jar files creation files --- Key: HDFS-475 URL: https://issues.apache.org/jira/browse/HDFS-475 Project: Hadoop HDFS Issue Type: Sub-task Components: build Reporter: Konstantin Boudnik Assignee: Konstantin Boudnik Fix For: 0.21.0 Attachments: HDFS-475.patch, HDFS-475.patch Current implementation of the FI framework allows to mix faults into production classes, e.g. into build/ folder. Although the default probability level is set to zero it doesn't look clean and might potentially over complicate the build and release process. FI related targets are better be logically and physically separated, e.g. to put instrumented artifacts into a separate folder, say, build-fi/ -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (HDFS-462) Unit tests not working under Windows
[ https://issues.apache.org/jira/browse/HDFS-462?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Konstantin Shvachko updated HDFS-462: - Resolution: Fixed Hadoop Flags: [Reviewed] Status: Resolved (was: Patch Available) I just committed this. Thank you Jakob. Unit tests not working under Windows Key: HDFS-462 URL: https://issues.apache.org/jira/browse/HDFS-462 Project: Hadoop HDFS Issue Type: Bug Components: name-node, test Affects Versions: 0.21.0 Environment: Windows Reporter: Luca Telloli Assignee: Jakob Homan Fix For: 0.21.0 Attachments: HDFS-462.patch, TEST-org.apache.hadoop.hdfs.TestHdfs.txt Unit tests are failing on windows due to a problem with rename. The failing code is around line 520 in FSImage.java: {noformat} assert curDir.exists() : Current directory must exist.; assert !prevDir.exists() : prvious directory must not exist.; assert !tmpDir.exists() : prvious.tmp directory must not exist.; // rename current to tmp rename(curDir, tmpDir); // save new image if (!curDir.mkdir()) throw new IOException(Cannot create directory + curDir); {noformat} and seems related to some open file or directory -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (HDFS-446) Offline Image Viewer Ls visitor incorrectly says 'output file' instead of 'input file'
[ https://issues.apache.org/jira/browse/HDFS-446?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Konstantin Shvachko updated HDFS-446: - Component/s: tools test Affects Version/s: 0.21.0 Fix Version/s: 0.21.0 Hadoop Flags: [Reviewed] Offline Image Viewer Ls visitor incorrectly says 'output file' instead of 'input file' -- Key: HDFS-446 URL: https://issues.apache.org/jira/browse/HDFS-446 Project: Hadoop HDFS Issue Type: Bug Components: test, tools Affects Versions: 0.21.0 Reporter: Jakob Homan Assignee: Jakob Homan Priority: Minor Fix For: 0.21.0 Attachments: HDFS-446.patch, HDFS-446.patch In the {{finishAbnormally}} method of Ls Visitor it should be input ended unexpectedly, not output. Trivial documentation change. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Assigned: (HDFS-476) Create a fi test target.
[ https://issues.apache.org/jira/browse/HDFS-476?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Konstantin Boudnik reassigned HDFS-476: --- Assignee: Konstantin Boudnik Create a fi test target. Key: HDFS-476 URL: https://issues.apache.org/jira/browse/HDFS-476 Project: Hadoop HDFS Issue Type: Sub-task Components: build Reporter: Tsz Wo (Nicholas), SZE Assignee: Konstantin Boudnik Unit tests may be created with fi. Current we have to to ant injectfaults and then ant test -DTestFiXxx for running a particular fi unit test TestFiXxx. It will be easier if there is an ant target, say test-fi, such that it compiles and runs all fi tests. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (HDFS-475) Create a separate targets for fault injection related test and jar files creation files
[ https://issues.apache.org/jira/browse/HDFS-475?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12730624#action_12730624 ] Konstantin Boudnik commented on HDFS-475: - HDFS-476 will be fixed by this patch as well although in a slightly different manner. Now in order to run tests with injected faults one doesn't need to run 'injectfaults' target separately. It could be taken care off in a single command: {code} ant run-hdfs-test-fi -DTestFiXxx {code} I believe this is the essence of HDFS-476. Also, this new version of the patch provides new targets to create dev. and test jar files with included FI instrumentation. Please be advised, that Hudson's test-patch output can't be provided for this patch, because HDFS's Hudson doesn't run new targets right now. Create a separate targets for fault injection related test and jar files creation files --- Key: HDFS-475 URL: https://issues.apache.org/jira/browse/HDFS-475 Project: Hadoop HDFS Issue Type: Sub-task Components: build Reporter: Konstantin Boudnik Assignee: Konstantin Boudnik Fix For: 0.21.0 Attachments: HDFS-475.patch, HDFS-475.patch, HDFS-475.patch Current implementation of the FI framework allows to mix faults into production classes, e.g. into build/ folder. Although the default probability level is set to zero it doesn't look clean and might potentially over complicate the build and release process. FI related targets are better be logically and physically separated, e.g. to put instrumented artifacts into a separate folder, say, build-fi/ -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (HDFS-200) In HDFS, sync() not yet guarantees data available to the new readers
[ https://issues.apache.org/jira/browse/HDFS-200?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12730666#action_12730666 ] stack commented on HDFS-200: (Thanks for review Konstantin) In my last few test runs, NameNode has shut itself down with the below: {code} ... 009-07-14 00:17:46,586 DEBUG org.apache.hadoop.hdfs.StateChange: BLOCK* NameSystem.removeStoredBlock: blk_-9156287469566772234_2527 from XX.XX.XX.142:51010 2009-07-14 00:17:46,586 DEBUG org.apache.hadoop.hdfs.StateChange: BLOCK* NameSystem.removeStoredBlock: blk_-9181830129071396520_2355 from XX.XX.XX.142:51010 2009-07-14 00:17:46,586 DEBUG org.apache.hadoop.hdfs.StateChange: BLOCK* NameSystem.removeStoredBlock: blk_-9205119721509648294_2410 from XX.XX.XX.142:51010 2009-07-14 00:17:46,586 INFO org.apache.hadoop.hdfs.StateChange: BLOCK* NameSystem.processReport: block blk_-7011715647341740217_1 on XX.XX.XX.142:51010 size 47027149 does not belong to any file. 2009-07-14 00:17:46,586 INFO org.apache.hadoop.hdfs.StateChange: BLOCK* NameSystem.addToInvalidates: blk_-7011715647341740217 is added to invalidSet of XX.XX.XX.142:51010 2009-07-14 00:17:46,586 INFO org.apache.hadoop.hdfs.StateChange: BLOCK* NameSystem.processReport: block blk_-280166356715716926_1 on XX.XX.XX.142:51010 size 6487 does not belong to any file. 2009-07-14 00:17:46,586 INFO org.apache.hadoop.hdfs.StateChange: BLOCK* NameSystem.addToInvalidates: blk_-280166356715716926 is added to invalidSet of XX.XX.XX.142:51010 2009-07-14 00:17:46,586 INFO org.apache.hadoop.hdfs.StateChange: BLOCK* NameSystem.processReport: block blk_1532053033915429278_1 on XX.XX.XX.142:51010 size 3869 does not belong to any file. 2009-07-14 00:17:46,586 INFO org.apache.hadoop.hdfs.StateChange: BLOCK* NameSystem.addToInvalidates: blk_1532053033915429278 is added to invalidSet of XX.XX.XX.142:51010 2009-07-14 00:17:47,303 WARN org.apache.hadoop.hdfs.server.namenode.FSNamesystem: ReplicationMonitor thread received Runtime exception. java.lang.IllegalStateException: generationStamp (=1) == GenerationStamp.WILDCARD_STAMP 2009-07-14 00:17:47,304 INFO org.apache.hadoop.hdfs.server.namenode.NameNode: SHUTDOWN_MSG: / SHUTDOWN_MSG: Shutting down NameNode at aa0-000-12.u.powerset.com/XX.XX.XX.139 / {code} My guess this is a bug only fellas with dfs.support.append=true set run in to? Here is code from ReplicationMonitor: {code} } catch (Throwable t) { LOG.warn(ReplicationMonitor thread received Runtime exception. + t); Runtime.getRuntime().exit(-1); } {code} Thats a rough call I'd say? There are no more detailed exceptions in NN log. Dig in more and stick what I find in another issue? In HDFS, sync() not yet guarantees data available to the new readers Key: HDFS-200 URL: https://issues.apache.org/jira/browse/HDFS-200 Project: Hadoop HDFS Issue Type: New Feature Reporter: Tsz Wo (Nicholas), SZE Assignee: dhruba borthakur Priority: Blocker Attachments: 4379_20081010TC3.java, fsyncConcurrentReaders.txt, fsyncConcurrentReaders11_20.txt, fsyncConcurrentReaders3.patch, fsyncConcurrentReaders4.patch, fsyncConcurrentReaders5.txt, fsyncConcurrentReaders6.patch, fsyncConcurrentReaders9.patch, hadoop-stack-namenode-aa0-000-12.u.powerset.com.log.gz, hypertable-namenode.log.gz, namenode.log, namenode.log, Reader.java, Reader.java, reopen_test.sh, ReopenProblem.java, Writer.java, Writer.java In the append design doc (https://issues.apache.org/jira/secure/attachment/12370562/Appends.doc), it says * A reader is guaranteed to be able to read data that was 'flushed' before the reader opened the file However, this feature is not yet implemented. Note that the operation 'flushed' is now called sync. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.