Re: [VOTE] hive release candidate 0.4.1-rc0
Hi Min, What is zip? Which codec does it use? I think it's probably a problem of the codec. Can you try GzipCodec? Most probably that will work fine. Zheng On Mon, Nov 2, 2009 at 1:08 AM, Min Zhou coderp...@gmail.com wrote: If it returns more than 0 rows, that error will never happen. Thanks, Min On Mon, Nov 2, 2009 at 5:06 PM, Min Zhou coderp...@gmail.com wrote: No, it's zip. On Mon, Nov 2, 2009 at 4:03 PM, Zheng Shao zsh...@gmail.com wrote: Do you mean gzip codec? I think an empty gzip file should be 20 bytes. There might be some problem with the gzip codec (or native gzip codec) on your cluster. Can you check the log message of the map tasks whether it has a line called Successfully loaded native gzip lib? You can try any query that produces empty results - it should go through the same code path. Zheng On Sun, Nov 1, 2009 at 11:07 PM, Min Zhou coderp...@gmail.com wrote: we use zip codec in default. Some of the same lines were omitted from the error stack: at org.apache.hadoop.hive.ql.exec.FetchOperator.getNextRow(FetchOperator.java:272) Thanks, Min On Mon, Nov 2, 2009 at 2:57 PM, Zheng Shao zsh...@gmail.com wrote: Min, can you check the default compression codec in your hadoop conf? The 8-byte file must be a compressed file using the codec which represents 0-length file. It seems that codec was not able to decompress the stream. Zheng On Sun, Nov 1, 2009 at 10:49 PM, Min Zhou coderp...@gmail.com wrote: I think there may be a bug still in this release. hiveselect stuff_status from auctions where auction_id='2591238417' and pt='20091027'; auctions is a table partitioned by date, it stored as a textfile w/o compression. The query above should return 0 rows. but when hive.exec.compress.output=true, hive will crash with a StackOverflowError java.lang.StackOverflowError at java.lang.ref.FinalReference.init(FinalReference.java:16) at java.lang.ref.Finalizer.init(Finalizer.java:66) at java.lang.ref.Finalizer.register(Finalizer.java:72) at java.lang.Object.init(Object.java:20) at java.net.SocketImpl.init(SocketImpl.java:27) at java.net.PlainSocketImpl.init(PlainSocketImpl.java:90) at java.net.SocksSocketImpl.init(SocksSocketImpl.java:33) at java.net.Socket.setImpl(Socket.java:434) at java.net.Socket.init(Socket.java:68) at sun.nio.ch.SocketAdaptor.init(SocketAdaptor.java:50) at sun.nio.ch.SocketAdaptor.create(SocketAdaptor.java:55) at sun.nio.ch.SocketChannelImpl.socket(SocketChannelImpl.java:105) at org.apache.hadoop.net.StandardSocketFactory.createSocket(StandardSocketFactory.java:58) at org.apache.hadoop.hdfs.DFSClient$DFSInputStream.blockSeekTo(DFSClient.java:1540) at org.apache.hadoop.hdfs.DFSClient$DFSInputStream.read(DFSClient.java:1662) at java.io.DataInputStream.read(DataInputStream.java:132) at org.apache.hadoop.io.compress.DecompressorStream.getCompressedData(DecompressorStream.java:96) at org.apache.hadoop.io.compress.DecompressorStream.decompress(DecompressorStream.java:86) at org.apache.hadoop.io.compress.DecompressorStream.read(DecompressorStream.java:74) at java.io.InputStream.read(InputStream.java:85) at org.apache.hadoop.util.LineReader.backfill(LineReader.java:82) at org.apache.hadoop.util.LineReader.readLine(LineReader.java:112) at org.apache.hadoop.mapred.LineRecordReader.next(LineRecordReader.java:134) at org.apache.hadoop.mapred.LineRecordReader.next(LineRecordReader.java:39) at org.apache.hadoop.hive.ql.exec.FetchOperator.getNextRow(FetchOperator.java:256) at org.apache.hadoop.hive.ql.exec.FetchOperator.getNextRow(FetchOperator.java:272) Each mapper will produce a 8 bytes deflate file on hdfs(we set hive.merge.mapfiles=false), their hex representation is like below: 78 9C 03 00 00 00 00 01 This is the reason why FetchOperator:272 is called recursively, and caused a stack overflow error. Regards, Min On Mon, Nov 2, 2009 at 6:34 AM, Zheng Shao zsh...@gmail.com wrote: I have made a release candidate 0.4.1-rc0. We've fixed several critical bugs to hive release 0.4.0. We need hive release 0.4.1 out asap. Here are the list of changes: HIVE-884. Metastore Server should call System.exit() on error. (Zheng Shao via pchakka) HIVE-864. Fix map-join memory-leak. (Namit Jain via zshao) HIVE-878. Update the hash table entry before flushing in Group By hash aggregation (Zheng Shao via namit) HIVE-882. Create a new directory every time for scratch. (Namit Jain via zshao) HIVE-890. Fix cli.sh for detecting Hadoop versions. (Paul Huff via zshao) HIVE-892. Hive to kill hadoop jobs using POST. (Dhruba Borthakur via zshao) HIVE-883. URISyntaxException
[VOTE] hive release candidate 0.4.1-rc0
I have made a release candidate 0.4.1-rc0. We've fixed several critical bugs to hive release 0.4.0. We need hive release 0.4.1 out asap. Here are the list of changes: HIVE-884. Metastore Server should call System.exit() on error. (Zheng Shao via pchakka) HIVE-864. Fix map-join memory-leak. (Namit Jain via zshao) HIVE-878. Update the hash table entry before flushing in Group By hash aggregation (Zheng Shao via namit) HIVE-882. Create a new directory every time for scratch. (Namit Jain via zshao) HIVE-890. Fix cli.sh for detecting Hadoop versions. (Paul Huff via zshao) HIVE-892. Hive to kill hadoop jobs using POST. (Dhruba Borthakur via zshao) HIVE-883. URISyntaxException when partition value contains special chars. (Zheng Shao via namit) Please vote. -- Yours, Zheng
Re: [VOTE] hive release candidate 0.4.1-rc0
I think there may be a bug still in this release. hiveselect stuff_status from auctions where auction_id='2591238417' and pt='20091027'; auctions is a table partitioned by date, it stored as a textfile w/o compression. The query above should return 0 rows. but when hive.exec.compress.output=true, hive will crash with a StackOverflowError java.lang.StackOverflowError at java.lang.ref.FinalReference.init(FinalReference.java:16) at java.lang.ref.Finalizer.init(Finalizer.java:66) at java.lang.ref.Finalizer.register(Finalizer.java:72) at java.lang.Object.init(Object.java:20) at java.net.SocketImpl.init(SocketImpl.java:27) at java.net.PlainSocketImpl.init(PlainSocketImpl.java:90) at java.net.SocksSocketImpl.init(SocksSocketImpl.java:33) at java.net.Socket.setImpl(Socket.java:434) at java.net.Socket.init(Socket.java:68) at sun.nio.ch.SocketAdaptor.init(SocketAdaptor.java:50) at sun.nio.ch.SocketAdaptor.create(SocketAdaptor.java:55) at sun.nio.ch.SocketChannelImpl.socket(SocketChannelImpl.java:105) at org.apache.hadoop.net.StandardSocketFactory.createSocket(StandardSocketFactory.java:58) at org.apache.hadoop.hdfs.DFSClient$DFSInputStream.blockSeekTo(DFSClient.java:1540) at org.apache.hadoop.hdfs.DFSClient$DFSInputStream.read(DFSClient.java:1662) at java.io.DataInputStream.read(DataInputStream.java:132) at org.apache.hadoop.io.compress.DecompressorStream.getCompressedData(DecompressorStream.java:96) at org.apache.hadoop.io.compress.DecompressorStream.decompress(DecompressorStream.java:86) at org.apache.hadoop.io.compress.DecompressorStream.read(DecompressorStream.java:74) at java.io.InputStream.read(InputStream.java:85) at org.apache.hadoop.util.LineReader.backfill(LineReader.java:82) at org.apache.hadoop.util.LineReader.readLine(LineReader.java:112) at org.apache.hadoop.mapred.LineRecordReader.next(LineRecordReader.java:134) at org.apache.hadoop.mapred.LineRecordReader.next(LineRecordReader.java:39) at org.apache.hadoop.hive.ql.exec.FetchOperator.getNextRow(FetchOperator.java:256) at org.apache.hadoop.hive.ql.exec.FetchOperator.getNextRow(FetchOperator.java:272) Each mapper will produce a 8 bytes deflate file on hdfs(we set hive.merge.mapfiles=false), their hex representation is like below: 78 9C 03 00 00 00 00 01 This is the reason why FetchOperator:272 is called recursively, and caused a stack overflow error. Regards, Min On Mon, Nov 2, 2009 at 6:34 AM, Zheng Shao zsh...@gmail.com wrote: I have made a release candidate 0.4.1-rc0. We've fixed several critical bugs to hive release 0.4.0. We need hive release 0.4.1 out asap. Here are the list of changes: HIVE-884. Metastore Server should call System.exit() on error. (Zheng Shao via pchakka) HIVE-864. Fix map-join memory-leak. (Namit Jain via zshao) HIVE-878. Update the hash table entry before flushing in Group By hash aggregation (Zheng Shao via namit) HIVE-882. Create a new directory every time for scratch. (Namit Jain via zshao) HIVE-890. Fix cli.sh for detecting Hadoop versions. (Paul Huff via zshao) HIVE-892. Hive to kill hadoop jobs using POST. (Dhruba Borthakur via zshao) HIVE-883. URISyntaxException when partition value contains special chars. (Zheng Shao via namit) Please vote. -- Yours, Zheng -- My research interests are distributed systems, parallel computing and bytecode based virtual machine. My profile: http://www.linkedin.com/in/coderplay My blog: http://coderplay.javaeye.com
Re: [VOTE] hive release candidate 0.4.1-rc0
Min, can you check the default compression codec in your hadoop conf? The 8-byte file must be a compressed file using the codec which represents 0-length file. It seems that codec was not able to decompress the stream. Zheng On Sun, Nov 1, 2009 at 10:49 PM, Min Zhou coderp...@gmail.com wrote: I think there may be a bug still in this release. hiveselect stuff_status from auctions where auction_id='2591238417' and pt='20091027'; auctions is a table partitioned by date, it stored as a textfile w/o compression. The query above should return 0 rows. but when hive.exec.compress.output=true, hive will crash with a StackOverflowError java.lang.StackOverflowError at java.lang.ref.FinalReference.init(FinalReference.java:16) at java.lang.ref.Finalizer.init(Finalizer.java:66) at java.lang.ref.Finalizer.register(Finalizer.java:72) at java.lang.Object.init(Object.java:20) at java.net.SocketImpl.init(SocketImpl.java:27) at java.net.PlainSocketImpl.init(PlainSocketImpl.java:90) at java.net.SocksSocketImpl.init(SocksSocketImpl.java:33) at java.net.Socket.setImpl(Socket.java:434) at java.net.Socket.init(Socket.java:68) at sun.nio.ch.SocketAdaptor.init(SocketAdaptor.java:50) at sun.nio.ch.SocketAdaptor.create(SocketAdaptor.java:55) at sun.nio.ch.SocketChannelImpl.socket(SocketChannelImpl.java:105) at org.apache.hadoop.net.StandardSocketFactory.createSocket(StandardSocketFactory.java:58) at org.apache.hadoop.hdfs.DFSClient$DFSInputStream.blockSeekTo(DFSClient.java:1540) at org.apache.hadoop.hdfs.DFSClient$DFSInputStream.read(DFSClient.java:1662) at java.io.DataInputStream.read(DataInputStream.java:132) at org.apache.hadoop.io.compress.DecompressorStream.getCompressedData(DecompressorStream.java:96) at org.apache.hadoop.io.compress.DecompressorStream.decompress(DecompressorStream.java:86) at org.apache.hadoop.io.compress.DecompressorStream.read(DecompressorStream.java:74) at java.io.InputStream.read(InputStream.java:85) at org.apache.hadoop.util.LineReader.backfill(LineReader.java:82) at org.apache.hadoop.util.LineReader.readLine(LineReader.java:112) at org.apache.hadoop.mapred.LineRecordReader.next(LineRecordReader.java:134) at org.apache.hadoop.mapred.LineRecordReader.next(LineRecordReader.java:39) at org.apache.hadoop.hive.ql.exec.FetchOperator.getNextRow(FetchOperator.java:256) at org.apache.hadoop.hive.ql.exec.FetchOperator.getNextRow(FetchOperator.java:272) Each mapper will produce a 8 bytes deflate file on hdfs(we set hive.merge.mapfiles=false), their hex representation is like below: 78 9C 03 00 00 00 00 01 This is the reason why FetchOperator:272 is called recursively, and caused a stack overflow error. Regards, Min On Mon, Nov 2, 2009 at 6:34 AM, Zheng Shao zsh...@gmail.com wrote: I have made a release candidate 0.4.1-rc0. We've fixed several critical bugs to hive release 0.4.0. We need hive release 0.4.1 out asap. Here are the list of changes: HIVE-884. Metastore Server should call System.exit() on error. (Zheng Shao via pchakka) HIVE-864. Fix map-join memory-leak. (Namit Jain via zshao) HIVE-878. Update the hash table entry before flushing in Group By hash aggregation (Zheng Shao via namit) HIVE-882. Create a new directory every time for scratch. (Namit Jain via zshao) HIVE-890. Fix cli.sh for detecting Hadoop versions. (Paul Huff via zshao) HIVE-892. Hive to kill hadoop jobs using POST. (Dhruba Borthakur via zshao) HIVE-883. URISyntaxException when partition value contains special chars. (Zheng Shao via namit) Please vote. -- Yours, Zheng -- My research interests are distributed systems, parallel computing and bytecode based virtual machine. My profile: http://www.linkedin.com/in/coderplay My blog: http://coderplay.javaeye.com -- Yours, Zheng
Re: [VOTE] hive release candidate 0.4.1-rc0
we use zip codec in default. Some of the same lines were omitted from the error stack: at org.apache.hadoop.hive.ql.exec.FetchOperator.getNextRow(FetchOperator.java:272) Thanks, Min On Mon, Nov 2, 2009 at 2:57 PM, Zheng Shao zsh...@gmail.com wrote: Min, can you check the default compression codec in your hadoop conf? The 8-byte file must be a compressed file using the codec which represents 0-length file. It seems that codec was not able to decompress the stream. Zheng On Sun, Nov 1, 2009 at 10:49 PM, Min Zhou coderp...@gmail.com wrote: I think there may be a bug still in this release. hiveselect stuff_status from auctions where auction_id='2591238417' and pt='20091027'; auctions is a table partitioned by date, it stored as a textfile w/o compression. The query above should return 0 rows. but when hive.exec.compress.output=true, hive will crash with a StackOverflowError java.lang.StackOverflowError at java.lang.ref.FinalReference.init(FinalReference.java:16) at java.lang.ref.Finalizer.init(Finalizer.java:66) at java.lang.ref.Finalizer.register(Finalizer.java:72) at java.lang.Object.init(Object.java:20) at java.net.SocketImpl.init(SocketImpl.java:27) at java.net.PlainSocketImpl.init(PlainSocketImpl.java:90) at java.net.SocksSocketImpl.init(SocksSocketImpl.java:33) at java.net.Socket.setImpl(Socket.java:434) at java.net.Socket.init(Socket.java:68) at sun.nio.ch.SocketAdaptor.init(SocketAdaptor.java:50) at sun.nio.ch.SocketAdaptor.create(SocketAdaptor.java:55) at sun.nio.ch.SocketChannelImpl.socket(SocketChannelImpl.java:105) at org.apache.hadoop.net.StandardSocketFactory.createSocket(StandardSocketFactory.java:58) at org.apache.hadoop.hdfs.DFSClient$DFSInputStream.blockSeekTo(DFSClient.java:1540) at org.apache.hadoop.hdfs.DFSClient$DFSInputStream.read(DFSClient.java:1662) at java.io.DataInputStream.read(DataInputStream.java:132) at org.apache.hadoop.io.compress.DecompressorStream.getCompressedData(DecompressorStream.java:96) at org.apache.hadoop.io.compress.DecompressorStream.decompress(DecompressorStream.java:86) at org.apache.hadoop.io.compress.DecompressorStream.read(DecompressorStream.java:74) at java.io.InputStream.read(InputStream.java:85) at org.apache.hadoop.util.LineReader.backfill(LineReader.java:82) at org.apache.hadoop.util.LineReader.readLine(LineReader.java:112) at org.apache.hadoop.mapred.LineRecordReader.next(LineRecordReader.java:134) at org.apache.hadoop.mapred.LineRecordReader.next(LineRecordReader.java:39) at org.apache.hadoop.hive.ql.exec.FetchOperator.getNextRow(FetchOperator.java:256) at org.apache.hadoop.hive.ql.exec.FetchOperator.getNextRow(FetchOperator.java:272) Each mapper will produce a 8 bytes deflate file on hdfs(we set hive.merge.mapfiles=false), their hex representation is like below: 78 9C 03 00 00 00 00 01 This is the reason why FetchOperator:272 is called recursively, and caused a stack overflow error. Regards, Min On Mon, Nov 2, 2009 at 6:34 AM, Zheng Shao zsh...@gmail.com wrote: I have made a release candidate 0.4.1-rc0. We've fixed several critical bugs to hive release 0.4.0. We need hive release 0.4.1 out asap. Here are the list of changes: HIVE-884. Metastore Server should call System.exit() on error. (Zheng Shao via pchakka) HIVE-864. Fix map-join memory-leak. (Namit Jain via zshao) HIVE-878. Update the hash table entry before flushing in Group By hash aggregation (Zheng Shao via namit) HIVE-882. Create a new directory every time for scratch. (Namit Jain via zshao) HIVE-890. Fix cli.sh for detecting Hadoop versions. (Paul Huff via zshao) HIVE-892. Hive to kill hadoop jobs using POST. (Dhruba Borthakur via zshao) HIVE-883. URISyntaxException when partition value contains special chars. (Zheng Shao via namit) Please vote. -- Yours, Zheng -- My research interests are distributed systems, parallel computing and bytecode based virtual machine. My profile: http://www.linkedin.com/in/coderplay My blog: http://coderplay.javaeye.com -- Yours, Zheng -- My research interests are distributed systems, parallel computing and bytecode based virtual machine. My profile: http://www.linkedin.com/in/coderplay My blog: http://coderplay.javaeye.com