[jira] [Created] (HADOOP-15453) hadoop fs -count -v report "-count: Illegal option -v"
zhoutai.zt created HADOOP-15453: --- Summary: hadoop fs -count -v report "-count: Illegal option -v" Key: HADOOP-15453 URL: https://issues.apache.org/jira/browse/HADOOP-15453 Project: Hadoop Common Issue Type: Bug Components: fs Affects Versions: 2.7.2 Reporter: zhoutai.zt [hadoop@Hadoop1 bin]$ ./hadoop fs -count -q -h -v SparkHis* -count: Illegal option -v Reading the source code, can't find the -v option. {code:java} // code placeholder private static final String OPTION_QUOTA = "q"; private static final String OPTION_HUMAN = "h"; public static final String NAME = "count"; public static final String USAGE = "[-" + OPTION_QUOTA + "] [-" + OPTION_HUMAN + "] ..."; {code} BUT the document states the -v option. [http://hadoop.apache.org/docs/r2.7.2/hadoop-project-dist/hadoop-common/FileSystemShell.html#count] -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-issues-h...@hadoop.apache.org
[jira] [Commented] (HADOOP-15109) TestDFSIO -read -random doesn't work on file sized 4GB
[ https://issues.apache.org/jira/browse/HADOOP-15109?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16288659#comment-16288659 ] zhoutai.zt commented on HADOOP-15109: - Thanks [~ajayydv]. The second patch looks good to me, +1. > TestDFSIO -read -random doesn't work on file sized 4GB > -- > > Key: HADOOP-15109 > URL: https://issues.apache.org/jira/browse/HADOOP-15109 > Project: Hadoop Common > Issue Type: Bug > Components: fs, test >Affects Versions: 3.0.0-beta1 >Reporter: zhoutai.zt >Assignee: Ajay Kumar >Priority: Minor > Attachments: HADOOP-15109.001.patch, HADOOP-15109.002.patch, Screen > Shot 2017-12-11 at 3.17.22 PM.png > > > TestDFSIO -read -random throws IllegalArgumentException on 4GB file. The > cause is: > {code:java} > private long nextOffset(long current) { > if(skipSize == 0) > return rnd.nextInt((int)(fileSize)); > if(skipSize > 0) > return (current < 0) ? 0 : (current + bufferSize + skipSize); > // skipSize < 0 > return (current < 0) ? Math.max(0, fileSize - bufferSize) : > Math.max(0, current + skipSize); > } > } > {code} > When {color:#d04437}_filesize_{color} exceeds signed int, (int)(filesize) > will be negative and cause Random.nextInt throws IllegalArgumentException("n > must be positive"). -- This message was sent by Atlassian JIRA (v6.4.14#64029) - To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-issues-h...@hadoop.apache.org
[jira] [Commented] (HADOOP-15109) TestDFSIO -read -random doesn't work on file sized 4GB
[ https://issues.apache.org/jira/browse/HADOOP-15109?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16286947#comment-16286947 ] zhoutai.zt commented on HADOOP-15109: - Thanks [~ajayydv]. Another way to generate a bounded random long. {code:java} ThreadLocalRandom.current().nextLong(fileSize) {code} > TestDFSIO -read -random doesn't work on file sized 4GB > -- > > Key: HADOOP-15109 > URL: https://issues.apache.org/jira/browse/HADOOP-15109 > Project: Hadoop Common > Issue Type: Bug > Components: fs, test >Affects Versions: 3.0.0-beta1 >Reporter: zhoutai.zt >Assignee: Ajay Kumar >Priority: Minor > Attachments: HADOOP-15109.001.patch, Screen Shot 2017-12-11 at > 3.17.22 PM.png > > > TestDFSIO -read -random throws IllegalArgumentException on 4GB file. The > cause is: > {code:java} > private long nextOffset(long current) { > if(skipSize == 0) > return rnd.nextInt((int)(fileSize)); > if(skipSize > 0) > return (current < 0) ? 0 : (current + bufferSize + skipSize); > // skipSize < 0 > return (current < 0) ? Math.max(0, fileSize - bufferSize) : > Math.max(0, current + skipSize); > } > } > {code} > When {color:#d04437}_filesize_{color} exceeds signed int, (int)(filesize) > will be negative and cause Random.nextInt throws IllegalArgumentException("n > must be positive"). -- This message was sent by Atlassian JIRA (v6.4.14#64029) - To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-issues-h...@hadoop.apache.org
[jira] [Updated] (HADOOP-15109) TestDFSIO -read -random doesn't work on file sized 4GB
[ https://issues.apache.org/jira/browse/HADOOP-15109?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] zhoutai.zt updated HADOOP-15109: Status: Open (was: Patch Available) > TestDFSIO -read -random doesn't work on file sized 4GB > -- > > Key: HADOOP-15109 > URL: https://issues.apache.org/jira/browse/HADOOP-15109 > Project: Hadoop Common > Issue Type: Bug > Components: fs, test >Affects Versions: 3.0.0-beta1 >Reporter: zhoutai.zt >Priority: Minor > > TestDFSIO -read -random throws IllegalArgumentException on 4GB file. The > cause is: > {code:java} > private long nextOffset(long current) { > if(skipSize == 0) > return rnd.nextInt((int)(fileSize)); > if(skipSize > 0) > return (current < 0) ? 0 : (current + bufferSize + skipSize); > // skipSize < 0 > return (current < 0) ? Math.max(0, fileSize - bufferSize) : > Math.max(0, current + skipSize); > } > } > {code} > When {color:#d04437}_filesize_{color} exceeds signed int, (int)(filesize) > will be negative and cause Random.nextInt throws IllegalArgumentException("n > must be positive"). -- This message was sent by Atlassian JIRA (v6.4.14#64029) - To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-issues-h...@hadoop.apache.org
[jira] [Updated] (HADOOP-15109) TestDFSIO -read -random doesn't work on file sized 4GB
[ https://issues.apache.org/jira/browse/HADOOP-15109?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] zhoutai.zt updated HADOOP-15109: Status: Patch Available (was: Open) > TestDFSIO -read -random doesn't work on file sized 4GB > -- > > Key: HADOOP-15109 > URL: https://issues.apache.org/jira/browse/HADOOP-15109 > Project: Hadoop Common > Issue Type: Bug > Components: fs, test >Affects Versions: 3.0.0-beta1 >Reporter: zhoutai.zt >Priority: Minor > > TestDFSIO -read -random throws IllegalArgumentException on 4GB file. The > cause is: > {code:java} > private long nextOffset(long current) { > if(skipSize == 0) > return rnd.nextInt((int)(fileSize)); > if(skipSize > 0) > return (current < 0) ? 0 : (current + bufferSize + skipSize); > // skipSize < 0 > return (current < 0) ? Math.max(0, fileSize - bufferSize) : > Math.max(0, current + skipSize); > } > } > {code} > When {color:#d04437}_filesize_{color} exceeds signed int, (int)(filesize) > will be negative and cause Random.nextInt throws IllegalArgumentException("n > must be positive"). -- This message was sent by Atlassian JIRA (v6.4.14#64029) - To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-issues-h...@hadoop.apache.org
[jira] [Updated] (HADOOP-15109) TestDFSIO -read -random doesn't work on file sized 4GB
[ https://issues.apache.org/jira/browse/HADOOP-15109?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] zhoutai.zt updated HADOOP-15109: Priority: Minor (was: Major) > TestDFSIO -read -random doesn't work on file sized 4GB > -- > > Key: HADOOP-15109 > URL: https://issues.apache.org/jira/browse/HADOOP-15109 > Project: Hadoop Common > Issue Type: Bug > Components: fs, test >Affects Versions: 3.0.0-beta1 >Reporter: zhoutai.zt >Priority: Minor > > TestDFSIO -read -random throws IllegalArgumentException on 4GB file. The > cause is: > {code:java} > private long nextOffset(long current) { > if(skipSize == 0) > return rnd.nextInt((int)(fileSize)); > if(skipSize > 0) > return (current < 0) ? 0 : (current + bufferSize + skipSize); > // skipSize < 0 > return (current < 0) ? Math.max(0, fileSize - bufferSize) : > Math.max(0, current + skipSize); > } > } > {code} > When {color:#d04437}_filesize_{color} exceeds signed int, (int)(filesize) > will be negative and cause Random.nextInt throws IllegalArgumentException("n > must be positive"). -- This message was sent by Atlassian JIRA (v6.4.14#64029) - To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-issues-h...@hadoop.apache.org
[jira] [Created] (HADOOP-15109) TestDFSIO -read -random doesn't work on file sized 4GB
zhoutai.zt created HADOOP-15109: --- Summary: TestDFSIO -read -random doesn't work on file sized 4GB Key: HADOOP-15109 URL: https://issues.apache.org/jira/browse/HADOOP-15109 Project: Hadoop Common Issue Type: Bug Components: fs, test Affects Versions: 3.0.0-beta1 Reporter: zhoutai.zt TestDFSIO -read -random throws IllegalArgumentException on 4GB file. The cause is: {code:java} private long nextOffset(long current) { if(skipSize == 0) return rnd.nextInt((int)(fileSize)); if(skipSize > 0) return (current < 0) ? 0 : (current + bufferSize + skipSize); // skipSize < 0 return (current < 0) ? Math.max(0, fileSize - bufferSize) : Math.max(0, current + skipSize); } } {code} When {color:#d04437}_filesize_{color} exceeds signed int, (int)(filesize) will be negative and cause Random.nextInt throws IllegalArgumentException("n must be positive"). -- This message was sent by Atlassian JIRA (v6.4.14#64029) - To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-issues-h...@hadoop.apache.org
[jira] [Commented] (HADOOP-15101) what testListStatusFile verified not consistent with listStatus declaration in FileSystem
[ https://issues.apache.org/jira/browse/HADOOP-15101?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16284550#comment-16284550 ] zhoutai.zt commented on HADOOP-15101: - Thanks Steve Loughran. Will the details in filesystem.md be add to FileSystem.java? At least add a link to filesystem.md. > what testListStatusFile verified not consistent with listStatus declaration > in FileSystem > --- > > Key: HADOOP-15101 > URL: https://issues.apache.org/jira/browse/HADOOP-15101 > Project: Hadoop Common > Issue Type: Bug > Components: fs, test >Affects Versions: 3.0.0-beta1 >Reporter: zhoutai.zt >Priority: Critical > > {code} > @Test > public void testListStatusFile() throws Throwable { > describe("test the listStatus(path) on a file"); > Path f = touchf("liststatusfile"); > verifyStatusArrayMatchesFile(f, getFileSystem().listStatus(f)); > } > {code} > In this case, first create a file _f_, then listStatus on _f_,expect > listStatus returns an array of one FileStatus. But this is not consistent > with the declarations in FileSystem, i.e. > {code} > " > List the statuses of the files/directories in the given path if the path is a > directory. > Parameters: > f given path > Returns: > the statuses of the files/directories in the given patch > " > {code} > Which is the expected? The behave in fs contract test or in FileSystem? -- This message was sent by Atlassian JIRA (v6.4.14#64029) - To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-issues-h...@hadoop.apache.org
[jira] [Issue Comment Deleted] (HADOOP-15101) what testListStatusFile verified not consistent with listStatus declaration in FileSystem
[ https://issues.apache.org/jira/browse/HADOOP-15101?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] zhoutai.zt updated HADOOP-15101: Comment: was deleted (was: Where can I find the file filesystem.md?) > what testListStatusFile verified not consistent with listStatus declaration > in FileSystem > --- > > Key: HADOOP-15101 > URL: https://issues.apache.org/jira/browse/HADOOP-15101 > Project: Hadoop Common > Issue Type: Bug > Components: fs, test >Affects Versions: 3.0.0-beta1 >Reporter: zhoutai.zt >Priority: Critical > > {code} > @Test > public void testListStatusFile() throws Throwable { > describe("test the listStatus(path) on a file"); > Path f = touchf("liststatusfile"); > verifyStatusArrayMatchesFile(f, getFileSystem().listStatus(f)); > } > {code} > In this case, first create a file _f_, then listStatus on _f_,expect > listStatus returns an array of one FileStatus. But this is not consistent > with the declarations in FileSystem, i.e. > {code} > " > List the statuses of the files/directories in the given path if the path is a > directory. > Parameters: > f given path > Returns: > the statuses of the files/directories in the given patch > " > {code} > Which is the expected? The behave in fs contract test or in FileSystem? -- This message was sent by Atlassian JIRA (v6.4.14#64029) - To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-issues-h...@hadoop.apache.org
[jira] [Commented] (HADOOP-15101) what testListStatusFile verified not consistent with listStatus declaration in FileSystem
[ https://issues.apache.org/jira/browse/HADOOP-15101?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16284546#comment-16284546 ] zhoutai.zt commented on HADOOP-15101: - Where can I find the file filesystem.md? > what testListStatusFile verified not consistent with listStatus declaration > in FileSystem > --- > > Key: HADOOP-15101 > URL: https://issues.apache.org/jira/browse/HADOOP-15101 > Project: Hadoop Common > Issue Type: Bug > Components: fs, test >Affects Versions: 3.0.0-beta1 >Reporter: zhoutai.zt >Priority: Critical > > {code} > @Test > public void testListStatusFile() throws Throwable { > describe("test the listStatus(path) on a file"); > Path f = touchf("liststatusfile"); > verifyStatusArrayMatchesFile(f, getFileSystem().listStatus(f)); > } > {code} > In this case, first create a file _f_, then listStatus on _f_,expect > listStatus returns an array of one FileStatus. But this is not consistent > with the declarations in FileSystem, i.e. > {code} > " > List the statuses of the files/directories in the given path if the path is a > directory. > Parameters: > f given path > Returns: > the statuses of the files/directories in the given patch > " > {code} > Which is the expected? The behave in fs contract test or in FileSystem? -- This message was sent by Atlassian JIRA (v6.4.14#64029) - To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-issues-h...@hadoop.apache.org
[jira] [Updated] (HADOOP-15101) what testListStatusFile verified not consistent with listStatus declaration in FileSystem
[ https://issues.apache.org/jira/browse/HADOOP-15101?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] zhoutai.zt updated HADOOP-15101: Description: @Test public void testListStatusFile() throws Throwable { describe("test the listStatus(path) on a file"); Path f = touchf("liststatusfile"); verifyStatusArrayMatchesFile(f, getFileSystem().listStatus(f)); } In this case, first create a file _f_, then listStatus on _f_,expect listStatus returns an array of one FileStatus. But this is not consistent with the declarations in FileSystem, i.e. " List the statuses of the files/directories in the given path if the path is a directory. Parameters: f given path Returns: the statuses of the files/directories in the given patch " Which is the expected? The behave in fs contract test or in FileSystem? was: @Test public void testListStatusFile() throws Throwable { describe("test the listStatus(path) on a file"); Path f = touchf("liststatusfile"); verifyStatusArrayMatchesFile(f, getFileSystem().listStatus(f)); } In this case, first create a file _f_, then listStatus on _f_,expect listStatus returns an array of one FileStatus. But this is not consistent with the declarations in FileSystem, i.e. ??List the statuses of the files/directories in the given path if the path is a directory. Parameters: f given path Returns: the statuses of the files/directories in the given patch?? Which is the expected? The behave in fs contract test or in FileSystem? > what testListStatusFile verified not consistent with listStatus declaration > in FileSystem > --- > > Key: HADOOP-15101 > URL: https://issues.apache.org/jira/browse/HADOOP-15101 > Project: Hadoop Common > Issue Type: Bug > Components: fs, test >Affects Versions: 3.0.0-beta1 >Reporter: zhoutai.zt >Priority: Critical > > @Test > public void testListStatusFile() throws Throwable { > describe("test the listStatus(path) on a file"); > Path f = touchf("liststatusfile"); > verifyStatusArrayMatchesFile(f, getFileSystem().listStatus(f)); > } > In this case, first create a file _f_, then listStatus on _f_,expect > listStatus returns an array of one FileStatus. But this is not consistent > with the declarations in FileSystem, i.e. > " > List the statuses of the files/directories in the given path if the path is a > directory. > Parameters: > f given path > Returns: > the statuses of the files/directories in the given patch > " > Which is the expected? The behave in fs contract test or in FileSystem? -- This message was sent by Atlassian JIRA (v6.4.14#64029) - To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-issues-h...@hadoop.apache.org
[jira] [Created] (HADOOP-15101) what testListStatusFile verified not consistent with listStatus declaration in FileSystem
zhoutai.zt created HADOOP-15101: --- Summary: what testListStatusFile verified not consistent with listStatus declaration in FileSystem Key: HADOOP-15101 URL: https://issues.apache.org/jira/browse/HADOOP-15101 Project: Hadoop Common Issue Type: Bug Components: fs, test Affects Versions: 3.0.0-beta1 Reporter: zhoutai.zt Priority: Critical @Test public void testListStatusFile() throws Throwable { describe("test the listStatus(path) on a file"); Path f = touchf("liststatusfile"); verifyStatusArrayMatchesFile(f, getFileSystem().listStatus(f)); } In this case, first create a file _f_, then listStatus on _f_,expect listStatus returns an array of one FileStatus. But this is not consistent with the declarations in FileSystem, i.e. ??List the statuses of the files/directories in the given path if the path is a directory. Parameters: f given path Returns: the statuses of the files/directories in the given patch?? Which is the expected? The behave in fs contract test or in FileSystem? -- This message was sent by Atlassian JIRA (v6.4.14#64029) - To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-issues-h...@hadoop.apache.org
[jira] [Created] (HADOOP-15097) AbstractContractDeleteTest::testDeleteNonEmptyDirRecursive with misleading path
zhoutai.zt created HADOOP-15097: --- Summary: AbstractContractDeleteTest::testDeleteNonEmptyDirRecursive with misleading path Key: HADOOP-15097 URL: https://issues.apache.org/jira/browse/HADOOP-15097 Project: Hadoop Common Issue Type: Bug Components: fs, test Affects Versions: 3.0.0-beta1 Reporter: zhoutai.zt Priority: Minor @Test public void testDeleteNonEmptyDirRecursive() throws Throwable { Path path = path("{color:red}testDeleteNonEmptyDirNonRecursive{color}"); mkdirs(path); Path file = new Path(path, "childfile"); ContractTestUtils.writeTextFile(getFileSystem(), file, "goodbye, world", true); assertDeleted(path, true); assertPathDoesNotExist("not deleted", file); } change testDeleteNonEmptyDirNonRecursive to testDeleteNonEmptyDirRecursive -- This message was sent by Atlassian JIRA (v6.4.14#64029) - To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-issues-h...@hadoop.apache.org
[jira] [Commented] (HADOOP-15020) NNBench not support run more than one map task on the same host
[ https://issues.apache.org/jira/browse/HADOOP-15020?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16241994#comment-16241994 ] zhoutai.zt commented on HADOOP-15020: - thanks. > NNBench not support run more than one map task on the same host > --- > > Key: HADOOP-15020 > URL: https://issues.apache.org/jira/browse/HADOOP-15020 > Project: Hadoop Common > Issue Type: Improvement > Components: benchmarks >Affects Versions: 2.7.2 > Environment: Hadoop 2.7.2 >Reporter: zhoutai.zt >Priority: Minor > > When benchmark NameNode performance with NNBench, I start with pseudo > distributed deploy. Everything goes well with "-maps 1". BUT with -maps N > (n>1) and -operation create_write, many exceptions meet during the benchmark. > Hostname is part of the file path, which can differentiate hosts. With more > than two map tasks run on the same host, more than two map tasks may operate > on the same file, leading to exceptions. > 17/11/07 15:22:32 INFO hdfs.NNBench: RAW DATA: AL Total #1: > 84 > 17/11/07 15:22:32 INFO hdfs.NNBench: RAW DATA: AL Total #2: > 43 > 17/11/07 15:22:32 INFO hdfs.NNBench: RAW DATA: TPS Total (ms): > 2570 > 17/11/07 15:22:32 INFO hdfs.NNBench:RAW DATA: Longest Map Time (ms): > 814.0 > 17/11/07 15:22:32 INFO hdfs.NNBench:RAW DATA: Late maps: 0 > 17/11/07 15:22:32 INFO hdfs.NNBench: {color:red}RAW DATA: # of > exceptions: 3000{color} > 2017-11-07 14:54:08,082 INFO org.apache.hadoop.hdfs.NNBench: Exception > recorded in op: Create/Write/Close > 2017-11-07 14:54:08,082 INFO org.apache.hadoop.hdfs.NNBench: Exception > recorded in op: Create/Write/Close > 2017-11-07 14:54:08,083 INFO org.apache.hadoop.hdfs.NNBench: Exception > recorded in op: Create/Write/Close -- This message was sent by Atlassian JIRA (v6.4.14#64029) - To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-issues-h...@hadoop.apache.org
[jira] [Updated] (HADOOP-15020) NNBench not support run more than one map task on the same host
[ https://issues.apache.org/jira/browse/HADOOP-15020?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] zhoutai.zt updated HADOOP-15020: Issue Type: Bug (was: Improvement) > NNBench not support run more than one map task on the same host > --- > > Key: HADOOP-15020 > URL: https://issues.apache.org/jira/browse/HADOOP-15020 > Project: Hadoop Common > Issue Type: Bug > Components: benchmarks >Affects Versions: 2.7.2 > Environment: Hadoop 2.7.2 >Reporter: zhoutai.zt >Priority: Minor > > When benchmark NameNode performance with NNBench, I start with pseudo > distributed deploy. Everything goes well with "-maps 1". BUT with -maps N > (n>1) and -operation create_write, many exceptions meet during the benchmark. > Hostname is part of the file path, which can differentiate hosts. With more > than two map tasks run on the same host, more than two map tasks may operate > on the same file, leading to exceptions. > 17/11/07 15:22:32 INFO hdfs.NNBench: RAW DATA: AL Total #1: > 84 > 17/11/07 15:22:32 INFO hdfs.NNBench: RAW DATA: AL Total #2: > 43 > 17/11/07 15:22:32 INFO hdfs.NNBench: RAW DATA: TPS Total (ms): > 2570 > 17/11/07 15:22:32 INFO hdfs.NNBench:RAW DATA: Longest Map Time (ms): > 814.0 > 17/11/07 15:22:32 INFO hdfs.NNBench:RAW DATA: Late maps: 0 > 17/11/07 15:22:32 INFO hdfs.NNBench: {color:red}RAW DATA: # of > exceptions: 3000{color} > 2017-11-07 14:54:08,082 INFO org.apache.hadoop.hdfs.NNBench: Exception > recorded in op: Create/Write/Close > 2017-11-07 14:54:08,082 INFO org.apache.hadoop.hdfs.NNBench: Exception > recorded in op: Create/Write/Close > 2017-11-07 14:54:08,083 INFO org.apache.hadoop.hdfs.NNBench: Exception > recorded in op: Create/Write/Close -- This message was sent by Atlassian JIRA (v6.4.14#64029) - To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-issues-h...@hadoop.apache.org
[jira] [Updated] (HADOOP-15020) NNBench not support run more than one map task on the same host
[ https://issues.apache.org/jira/browse/HADOOP-15020?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] zhoutai.zt updated HADOOP-15020: Description: When benchmark NameNode performance with NNBench, I start with pseudo distributed deploy. Everything goes well with "-maps 1". BUT with -maps N (n>1) and -operation create_write, many exceptions meet during the benchmark. Hostname is part of the file path, which can differentiate hosts. With more than two map tasks run on the same host, more than two map tasks may operate on the same file, leading to exceptions. 17/11/07 15:22:32 INFO hdfs.NNBench: RAW DATA: AL Total #1: 84 17/11/07 15:22:32 INFO hdfs.NNBench: RAW DATA: AL Total #2: 43 17/11/07 15:22:32 INFO hdfs.NNBench: RAW DATA: TPS Total (ms): 2570 17/11/07 15:22:32 INFO hdfs.NNBench:RAW DATA: Longest Map Time (ms): 814.0 17/11/07 15:22:32 INFO hdfs.NNBench:RAW DATA: Late maps: 0 17/11/07 15:22:32 INFO hdfs.NNBench: {color:red}RAW DATA: # of exceptions: 3000{color} 2017-11-07 14:54:08,082 INFO org.apache.hadoop.hdfs.NNBench: Exception recorded in op: Create/Write/Close 2017-11-07 14:54:08,082 INFO org.apache.hadoop.hdfs.NNBench: Exception recorded in op: Create/Write/Close 2017-11-07 14:54:08,083 INFO org.apache.hadoop.hdfs.NNBench: Exception recorded in op: Create/Write/Close was: When benchmark NameNode performance with NNBench. I start with pseudo distributed deploy. Everything goes well with "-maps 1". BUT with -maps N (n>1) and -operation create_write, many exceptions meet during the benchmark. Hostname is part of the file path, which can differentiate hosts. With more than two map tasks run on the same host, more than two map tasks may operate on the same file, leading to exceptions. 17/11/07 15:22:32 INFO hdfs.NNBench: RAW DATA: AL Total #1: 84 17/11/07 15:22:32 INFO hdfs.NNBench: RAW DATA: AL Total #2: 43 17/11/07 15:22:32 INFO hdfs.NNBench: RAW DATA: TPS Total (ms): 2570 17/11/07 15:22:32 INFO hdfs.NNBench:RAW DATA: Longest Map Time (ms): 814.0 17/11/07 15:22:32 INFO hdfs.NNBench:RAW DATA: Late maps: 0 17/11/07 15:22:32 INFO hdfs.NNBench: {color:red}RAW DATA: # of exceptions: 3000{color} 2017-11-07 14:54:08,082 INFO org.apache.hadoop.hdfs.NNBench: Exception recorded in op: Create/Write/Close 2017-11-07 14:54:08,082 INFO org.apache.hadoop.hdfs.NNBench: Exception recorded in op: Create/Write/Close 2017-11-07 14:54:08,083 INFO org.apache.hadoop.hdfs.NNBench: Exception recorded in op: Create/Write/Close > NNBench not support run more than one map task on the same host > --- > > Key: HADOOP-15020 > URL: https://issues.apache.org/jira/browse/HADOOP-15020 > Project: Hadoop Common > Issue Type: Improvement > Components: benchmarks >Affects Versions: 2.7.2 > Environment: Hadoop 2.7.2 >Reporter: zhoutai.zt >Priority: Minor > > When benchmark NameNode performance with NNBench, I start with pseudo > distributed deploy. Everything goes well with "-maps 1". BUT with -maps N > (n>1) and -operation create_write, many exceptions meet during the benchmark. > Hostname is part of the file path, which can differentiate hosts. With more > than two map tasks run on the same host, more than two map tasks may operate > on the same file, leading to exceptions. > 17/11/07 15:22:32 INFO hdfs.NNBench: RAW DATA: AL Total #1: > 84 > 17/11/07 15:22:32 INFO hdfs.NNBench: RAW DATA: AL Total #2: > 43 > 17/11/07 15:22:32 INFO hdfs.NNBench: RAW DATA: TPS Total (ms): > 2570 > 17/11/07 15:22:32 INFO hdfs.NNBench:RAW DATA: Longest Map Time (ms): > 814.0 > 17/11/07 15:22:32 INFO hdfs.NNBench:RAW DATA: Late maps: 0 > 17/11/07 15:22:32 INFO hdfs.NNBench: {color:red}RAW DATA: # of > exceptions: 3000{color} > 2017-11-07 14:54:08,082 INFO org.apache.hadoop.hdfs.NNBench: Exception > recorded in op: Create/Write/Close > 2017-11-07 14:54:08,082 INFO org.apache.hadoop.hdfs.NNBench: Exception > recorded in op: Create/Write/Close > 2017-11-07 14:54:08,083 INFO org.apache.hadoop.hdfs.NNBench: Exception > recorded in op: Create/Write/Close -- This message was sent by Atlassian JIRA (v6.4.14#64029) - To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-issues-h...@hadoop.apache.org
[jira] [Updated] (HADOOP-15020) NNBench not support run more than one map task on the same host
[ https://issues.apache.org/jira/browse/HADOOP-15020?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] zhoutai.zt updated HADOOP-15020: Summary: NNBench not support run more than one map task on the same host (was: NNBench not support run more than one map task on one host) > NNBench not support run more than one map task on the same host > --- > > Key: HADOOP-15020 > URL: https://issues.apache.org/jira/browse/HADOOP-15020 > Project: Hadoop Common > Issue Type: Improvement > Components: benchmarks >Affects Versions: 2.7.2 > Environment: Hadoop 2.7.2 >Reporter: zhoutai.zt >Priority: Minor > > When benchmark NameNode performance with NNBench. I start with pseudo > distributed deploy. Everything goes well with "-maps 1". BUT with -maps N > (n>1) and -operation create_write, many exceptions meet during the benchmark. > Hostname is part of the file path, which can differentiate hosts. With more > than two map tasks run on the same host, more than two map tasks may operate > on the same file, leading to exceptions. > 17/11/07 15:22:32 INFO hdfs.NNBench: RAW DATA: AL Total #1: > 84 > 17/11/07 15:22:32 INFO hdfs.NNBench: RAW DATA: AL Total #2: > 43 > 17/11/07 15:22:32 INFO hdfs.NNBench: RAW DATA: TPS Total (ms): > 2570 > 17/11/07 15:22:32 INFO hdfs.NNBench:RAW DATA: Longest Map Time (ms): > 814.0 > 17/11/07 15:22:32 INFO hdfs.NNBench:RAW DATA: Late maps: 0 > 17/11/07 15:22:32 INFO hdfs.NNBench: {color:red}RAW DATA: # of > exceptions: 3000{color} > 2017-11-07 14:54:08,082 INFO org.apache.hadoop.hdfs.NNBench: Exception > recorded in op: Create/Write/Close > 2017-11-07 14:54:08,082 INFO org.apache.hadoop.hdfs.NNBench: Exception > recorded in op: Create/Write/Close > 2017-11-07 14:54:08,083 INFO org.apache.hadoop.hdfs.NNBench: Exception > recorded in op: Create/Write/Close -- This message was sent by Atlassian JIRA (v6.4.14#64029) - To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-issues-h...@hadoop.apache.org
[jira] [Created] (HADOOP-15020) NNBench not support run more than one map task on one host
zhoutai.zt created HADOOP-15020: --- Summary: NNBench not support run more than one map task on one host Key: HADOOP-15020 URL: https://issues.apache.org/jira/browse/HADOOP-15020 Project: Hadoop Common Issue Type: Improvement Components: benchmarks Affects Versions: 2.7.2 Environment: Hadoop 2.7.2 Reporter: zhoutai.zt Priority: Minor When benchmark NameNode performance with NNBench. I start with pseudo distributed deploy. Everything goes well with "-maps 1". BUT with -maps N (n>1) and -operation create_write, many exceptions meet during the benchmark. Hostname is part of the file path, which can differentiate hosts. With more than two map tasks run on the same host, more than two map tasks may operate on the same file, leading to exceptions. 17/11/07 15:22:32 INFO hdfs.NNBench: RAW DATA: AL Total #1: 84 17/11/07 15:22:32 INFO hdfs.NNBench: RAW DATA: AL Total #2: 43 17/11/07 15:22:32 INFO hdfs.NNBench: RAW DATA: TPS Total (ms): 2570 17/11/07 15:22:32 INFO hdfs.NNBench:RAW DATA: Longest Map Time (ms): 814.0 17/11/07 15:22:32 INFO hdfs.NNBench:RAW DATA: Late maps: 0 17/11/07 15:22:32 INFO hdfs.NNBench: {color:red}RAW DATA: # of exceptions: 3000{color} 2017-11-07 14:54:08,082 INFO org.apache.hadoop.hdfs.NNBench: Exception recorded in op: Create/Write/Close 2017-11-07 14:54:08,082 INFO org.apache.hadoop.hdfs.NNBench: Exception recorded in op: Create/Write/Close 2017-11-07 14:54:08,083 INFO org.apache.hadoop.hdfs.NNBench: Exception recorded in op: Create/Write/Close -- This message was sent by Atlassian JIRA (v6.4.14#64029) - To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-issues-h...@hadoop.apache.org