FAQ
Hi Rahul,

Moving this over to impala-user@cloudera.org, which should be able to help
you with this issue.


--
Aaron T. Myers
Software Engineer, Cloudera


On Thu, Apr 25, 2013 at 6:05 AM, rahul mathur
wrote:
Hi,

I am facing an error : "*Error communicating with impalad: TSocket read 0
bytes" *which causes a JRE crash, this error results when I run select
query on impala-shell. Following are the logs generated after this error in
impala_server.log :

13/04/25 11:09:23 INFO service.Frontend: analyze query select * from ex
13/04/25 11:09:23 INFO catalog.HdfsTable: load table ex
13/04/25 11:09:24 INFO catalog.HdfsTable: load partition block md for ex
13/04/25 11:09:24 INFO catalog.HdfsTable: loaded partition
PartitionBlockMetadata{#blocks=0, #filenames=0, totalStringLen=0}
13/04/25 11:09:24 INFO catalog.HdfsTable: loaded partition
PartitionBlockMetadata{#blocks=1, #filenames=1, totalStringLen=104}
13/04/25 11:09:24 INFO catalog.HdfsTable: loaded disk ids for table
default.ex
13/04/25 11:09:24 INFO catalog.HdfsTable: 2
13/04/25 11:09:24 INFO catalog.HdfsTable: table #rows=0
13/04/25 11:09:24 INFO service.Frontend: create plan
13/04/25 11:09:24 INFO planner.Planner: create single-node plan
13/04/25 11:09:24 INFO planner.HdfsScanNode: collecting partitions for
table ex
13/04/25 11:09:24 INFO planner.HdfsScanNode: finalize HdfsScan:
cardinality=0
13/04/25 11:09:24 INFO planner.HdfsScanNode: finalize HdfsScan: #nodes=2
13/04/25 11:09:24 INFO planner.Planner: create plan fragments
13/04/25 11:09:24 INFO planner.Planner: finalize plan fragments
13/04/25 11:09:25 INFO service.Frontend: get scan range locations
13/04/25 11:09:25 INFO service.Frontend: create result set metadata
13/04/25 11:09:25 INFO service.JniFrontend: Plan Fragment 0
UNPARTITIONED
EXCHANGE (1)
TUPLE IDS: 0

Plan Fragment 1
RANDOM
STREAM DATA SINK
EXCHANGE ID: 1
UNPARTITIONED

SCAN HDFS table=default.ex #partitions=1 size=265B (0)
TUPLE IDS: 0

#
# A fatal error has been detected by the Java Runtime Environment:
#
# SIGILL (0x4) at pc=0x00000000008ed857, pid=22887, tid=140607123736320
#
# JRE version: 7.0_21-b11
# Java VM: Java HotSpot(TM) 64-Bit Server VM (23.21-b01 mixed mode
linux-amd64 compressed oops)
# Problematic frame:
# C [impalad+0x4ed857] * impala::HdfsScanNode::UpdateCounters()+0x97*
#
# Failed to write core dump. Core dumps have been disabled. To enable core
dumping, try "ulimit -c unlimited" before starting Java again
#
# An error report file with more information is saved as:
# */var/run/impala/hs_err_pid22887.log*
#
# If you would like to submit a bug report, please visit:
# http://bugreport.sun.com/bugreport/crash.jsp


There is one more log file generated under the home directory of impala
user like "hs_err_pid22887.log" by JRE.
I tried running same queries on HIVE which showed correct result from
local and remote client. Also, tried both 0.7 and latest version 1.0 but
both are throwing same error. Please reply with a solution ASAP.

Thanks in advance,
*Rahul Mathur*

--


Search Discussions

  • 宾莉金 at Apr 26, 2013 at 12:45 pm
    I meet the same problem.
      Ubuntu 12.04.2 LTS




    2013/4/26 Lenni Kuff <lskuff@cloudera.com>
    Hi Rahaul,
    To confirm we have a fix that will work for you, could you please send
    us the hs_err_pid.log file from your crash? From the output you sent,
    this should be located here:

    /var/run/impala/hs_err_pid22887.log

    Thanks,
    Lenni
    Software Engineer - Cloudera
    On Thu, Apr 25, 2013 at 3:30 PM, Miklos Christine wrote:
    Hello Rahul,

    I believe this issue has been hit by others on the mailing list.
    See the mailing list for the subject "every impalad daemon crashes".

    There has been a jira filed for the issue and it will be fixed in the
    upcoming GA release, which will be the v1.0.
    Currently, I believe you are on v0.7.1 which still has the issue.
    https://issues.cloudera.org/browse/IMPALA-289

    Thanks,
    Miklos

    On Thu, Apr 25, 2013 at 1:58 PM, Aaron T. Myers wrote:

    Hi Rahul,

    Moving this over to impala-user@cloudera.org, which should be able to
    help
    you with this issue.


    --
    Aaron T. Myers
    Software Engineer, Cloudera


    On Thu, Apr 25, 2013 at 6:05 AM, rahul mathur
    wrote:
    Hi,

    I am facing an error : "Error communicating with impalad: TSocket read
    bytes" which causes a JRE crash, this error results when I run select
    query
    on impala-shell. Following are the logs generated after this error in
    impala_server.log :

    13/04/25 11:09:23 INFO service.Frontend: analyze query select * from ex
    13/04/25 11:09:23 INFO catalog.HdfsTable: load table ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: load partition block md for
    ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded partition
    PartitionBlockMetadata{#blocks=0, #filenames=0, totalStringLen=0}
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded partition
    PartitionBlockMetadata{#blocks=1, #filenames=1, totalStringLen=104}
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded disk ids for table
    default.ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: 2
    13/04/25 11:09:24 INFO catalog.HdfsTable: table #rows=0
    13/04/25 11:09:24 INFO service.Frontend: create plan
    13/04/25 11:09:24 INFO planner.Planner: create single-node plan
    13/04/25 11:09:24 INFO planner.HdfsScanNode: collecting partitions for
    table ex
    13/04/25 11:09:24 INFO planner.HdfsScanNode: finalize HdfsScan:
    cardinality=0
    13/04/25 11:09:24 INFO planner.HdfsScanNode: finalize HdfsScan:
    #nodes=2
    13/04/25 11:09:24 INFO planner.Planner: create plan fragments
    13/04/25 11:09:24 INFO planner.Planner: finalize plan fragments
    13/04/25 11:09:25 INFO service.Frontend: get scan range locations
    13/04/25 11:09:25 INFO service.Frontend: create result set metadata
    13/04/25 11:09:25 INFO service.JniFrontend: Plan Fragment 0
    UNPARTITIONED
    EXCHANGE (1)
    TUPLE IDS: 0

    Plan Fragment 1
    RANDOM
    STREAM DATA SINK
    EXCHANGE ID: 1
    UNPARTITIONED

    SCAN HDFS table=default.ex #partitions=1 size=265B (0)
    TUPLE IDS: 0

    #
    # A fatal error has been detected by the Java Runtime Environment:
    #
    # SIGILL (0x4) at pc=0x00000000008ed857, pid=22887,
    tid=140607123736320
    #
    # JRE version: 7.0_21-b11
    # Java VM: Java HotSpot(TM) 64-Bit Server VM (23.21-b01 mixed mode
    linux-amd64 compressed oops)
    # Problematic frame:
    # C [impalad+0x4ed857] impala::HdfsScanNode::UpdateCounters()+0x97
    #
    # Failed to write core dump. Core dumps have been disabled. To enable
    core dumping, try "ulimit -c unlimited" before starting Java again
    #
    # An error report file with more information is saved as:
    # /var/run/impala/hs_err_pid22887.log
    #
    # If you would like to submit a bug report, please visit:
    # http://bugreport.sun.com/bugreport/crash.jsp


    There is one more log file generated under the home directory of impala
    user like "hs_err_pid22887.log" by JRE.
    I tried running same queries on HIVE which showed correct result from
    local and remote client. Also, tried both 0.7 and latest version 1.0
    but
    both are throwing same error. Please reply with a solution ASAP.

    Thanks in advance,
    Rahul Mathur

    --



    --
    *Best Regards,*
      lijin bin
  • 宾莉金 at Apr 26, 2013 at 12:49 pm
    Impala download from
    http://beta.cloudera.com/impala/ubuntu/precise/amd64/impala/pool/contrib/i/impala/



    2013/4/26 宾莉金 <binlijin@gmail.com>
    I meet the same problem.
    Ubuntu 12.04.2 LTS




    2013/4/26 Lenni Kuff <lskuff@cloudera.com>
    Hi Rahaul,
    To confirm we have a fix that will work for you, could you please send
    us the hs_err_pid.log file from your crash? From the output you sent,
    this should be located here:

    /var/run/impala/hs_err_pid22887.log

    Thanks,
    Lenni
    Software Engineer - Cloudera

    On Thu, Apr 25, 2013 at 3:30 PM, Miklos Christine <mwc@cloudera.com>
    wrote:
    Hello Rahul,

    I believe this issue has been hit by others on the mailing list.
    See the mailing list for the subject "every impalad daemon crashes".

    There has been a jira filed for the issue and it will be fixed in the
    upcoming GA release, which will be the v1.0.
    Currently, I believe you are on v0.7.1 which still has the issue.
    https://issues.cloudera.org/browse/IMPALA-289

    Thanks,
    Miklos


    On Thu, Apr 25, 2013 at 1:58 PM, Aaron T. Myers <atm@cloudera.com>
    wrote:
    Hi Rahul,

    Moving this over to impala-user@cloudera.org, which should be able to
    help
    you with this issue.


    --
    Aaron T. Myers
    Software Engineer, Cloudera


    On Thu, Apr 25, 2013 at 6:05 AM, rahul mathur
    wrote:
    Hi,

    I am facing an error : "Error communicating with impalad: TSocket
    read 0
    bytes" which causes a JRE crash, this error results when I run select
    query
    on impala-shell. Following are the logs generated after this error in
    impala_server.log :

    13/04/25 11:09:23 INFO service.Frontend: analyze query select * from
    ex
    13/04/25 11:09:23 INFO catalog.HdfsTable: load table ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: load partition block md for
    ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded partition
    PartitionBlockMetadata{#blocks=0, #filenames=0, totalStringLen=0}
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded partition
    PartitionBlockMetadata{#blocks=1, #filenames=1, totalStringLen=104}
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded disk ids for table
    default.ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: 2
    13/04/25 11:09:24 INFO catalog.HdfsTable: table #rows=0
    13/04/25 11:09:24 INFO service.Frontend: create plan
    13/04/25 11:09:24 INFO planner.Planner: create single-node plan
    13/04/25 11:09:24 INFO planner.HdfsScanNode: collecting partitions for
    table ex
    13/04/25 11:09:24 INFO planner.HdfsScanNode: finalize HdfsScan:
    cardinality=0
    13/04/25 11:09:24 INFO planner.HdfsScanNode: finalize HdfsScan:
    #nodes=2
    13/04/25 11:09:24 INFO planner.Planner: create plan fragments
    13/04/25 11:09:24 INFO planner.Planner: finalize plan fragments
    13/04/25 11:09:25 INFO service.Frontend: get scan range locations
    13/04/25 11:09:25 INFO service.Frontend: create result set metadata
    13/04/25 11:09:25 INFO service.JniFrontend: Plan Fragment 0
    UNPARTITIONED
    EXCHANGE (1)
    TUPLE IDS: 0

    Plan Fragment 1
    RANDOM
    STREAM DATA SINK
    EXCHANGE ID: 1
    UNPARTITIONED

    SCAN HDFS table=default.ex #partitions=1 size=265B (0)
    TUPLE IDS: 0

    #
    # A fatal error has been detected by the Java Runtime Environment:
    #
    # SIGILL (0x4) at pc=0x00000000008ed857, pid=22887,
    tid=140607123736320
    #
    # JRE version: 7.0_21-b11
    # Java VM: Java HotSpot(TM) 64-Bit Server VM (23.21-b01 mixed mode
    linux-amd64 compressed oops)
    # Problematic frame:
    # C [impalad+0x4ed857] impala::HdfsScanNode::UpdateCounters()+0x97
    #
    # Failed to write core dump. Core dumps have been disabled. To enable
    core dumping, try "ulimit -c unlimited" before starting Java again
    #
    # An error report file with more information is saved as:
    # /var/run/impala/hs_err_pid22887.log
    #
    # If you would like to submit a bug report, please visit:
    # http://bugreport.sun.com/bugreport/crash.jsp


    There is one more log file generated under the home directory of
    impala
    user like "hs_err_pid22887.log" by JRE.
    I tried running same queries on HIVE which showed correct result from
    local and remote client. Also, tried both 0.7 and latest version 1.0
    but
    both are throwing same error. Please reply with a solution ASAP.

    Thanks in advance,
    Rahul Mathur

    --



    --
    *Best Regards,*
    lijin bin


    --
    *Best Regards,*
      lijin bin
  • 宾莉金 at Apr 26, 2013 at 1:55 pm
    The core dump is


    warning: Can't read pathname for load map: Input/output error.
    [Thread debugging using libthread_db enabled]
    Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
    Core was generated by
    `/home/tianzhao.blj/cloudera/impala/be/build/debug/service/impalad
    -state_store_'.
    Program terminated with signal 6, Aborted.
    #0 0x00007f35fd6fa425 in raise () from /lib/x86_64-linux-gnu/libc.so.6
    (gdb) bt
    #0 0x00007f35fd6fa425 in raise () from /lib/x86_64-linux-gnu/libc.so.6
    #1 0x00007f35fd6fdb8b in abort () from /lib/x86_64-linux-gnu/libc.so.6
    #2 0x00007f35fca8fa37 in os::abort(bool) () from
    /usr/lib/jvm/java-6-oracle/jre/lib/amd64/server/libjvm.so
    #3 0x00007f35fcbe3f28 in VMError::report_and_die() () from
    /usr/lib/jvm/java-6-oracle/jre/lib/amd64/server/libjvm.so
    #4 0x00007f35fca963f5 in JVM_handle_linux_signal () from
    /usr/lib/jvm/java-6-oracle/jre/lib/amd64/server/libjvm.so
    #5 0x00007f35fca926fe in signalHandler(int, siginfo*, void*) () from
    /usr/lib/jvm/java-6-oracle/jre/lib/amd64/server/libjvm.so
    #6 <signal handler called>
    #7 impala::HdfsScanNode::UpdateCounters (this=0x37e6800) at
    /home/tianzhao.blj/cloudera/impala/be/src/exec/hdfs-scan-node.cc:842
    #8 0x0000000000e5b9cc in impala::HdfsScanNode::GetNext (this=0x37e6800,
    state=0x51cca80, row_batch=0x45f4a00, eos=0x51c8171)
         at /home/tianzhao.blj/cloudera/impala/be/src/exec/hdfs-scan-node.cc:162
    #9 0x0000000000d0af2d in impala::PlanFragmentExecutor::GetNextInternal
    (this=0x51c8038, batch=0x7f35de6203b8)
         at
    /home/tianzhao.blj/cloudera/impala/be/src/runtime/plan-fragment-executor.cc:377
    #10 0x0000000000d09b0a in impala::PlanFragmentExecutor::OpenInternal
    (this=0x51c8038)
         at
    /home/tianzhao.blj/cloudera/impala/be/src/runtime/plan-fragment-executor.cc:249
    #11 0x0000000000d09885 in impala::PlanFragmentExecutor::Open
    (this=0x51c8038)
         at
    /home/tianzhao.blj/cloudera/impala/be/src/runtime/plan-fragment-executor.cc:226
    #12 0x0000000000b323a3 in impala::ImpalaServer::FragmentExecState::Exec
    (this=0x51c8000)
         at
    /home/tianzhao.blj/cloudera/impala/be/src/service/impala-server.cc:470
    #13 0x0000000000b45c67 in impala::ImpalaServer::RunExecPlanFragment
    (this=0x3cddc00, exec_state=0x51c8000)
         at
    /home/tianzhao.blj/cloudera/impala/be/src/service/impala-server.cc:1692
    #14 0x0000000000bad05f in boost::_mfi::mf1<void, impala::ImpalaServer,
    impala::ImpalaServer::FragmentExecState*>::operator() (
         this=0x2b1f308, p=0x3cddc00, a1=0x51c8000) at
    /usr/include/boost/bind/mem_fn_template.hpp:165
    #15 0x0000000000bac741 in
    boost::_bi::list2<boost::_bi::value<impala::ImpalaServer*>,
    boost::_bi::value<impala::ImpalaServer::FragmentExecState*>
    ::operator()<boost::_mfi::mf1<void, impala::ImpalaServer,
    impala::ImpalaServer::FragmentExecState*>, boost::_bi::list0>
    (this=0x2b1f318, f=..., a=...) at /usr/include/boost/bind/bind.hpp:313
    #16 0x0000000000bab717 in boost::_bi::bind_t<void, boost::_mfi::mf1<void,
    impala::ImpalaServer, impala::ImpalaServer::FragmentExecState*>,
    boost::_bi::list2<boost::_bi::value<impala::ImpalaServer*>,
    boost::_bi::value<impala::ImpalaServer::FragmentExecState*> > >::operator()
    (this=0x2b1f308) at /usr/include/boost/bind/bind_template.hpp:20
    #17 0x0000000000baa48a in
    boost::detail::thread_data<boost::_bi::bind_t<void, boost::_mfi::mf1<void,
    impala::ImpalaServer, impala::ImpalaServer::FragmentExecState*>,
    boost::_bi::list2<boost::_bi::value<impala::ImpalaServer*>,
    boost::_bi::value<impala::ImpalaServer::FragmentExecState*> > > >::run
    (this=0x2b1f180) at /usr/include/boost/thread/detail/thread.hpp:61
    #18 0x00007f35ff62ace9 in thread_proxy () from
    /usr/lib/libboost_thread.so.1.46.1
    #19 0x00007f35ff408e9a in start_thread () from
    /lib/x86_64-linux-gnu/libpthread.so.0
    #20 0x00007f35fd7b7ccd in clone () from /lib/x86_64-linux-gnu/libc.so.6
    #21 0x0000000000000000 in ?? ()



    java version "1.6.0_45"
    Java(TM) SE Runtime Environment (build 1.6.0_45-b06)
    Java HotSpot(TM) 64-Bit Server VM (build 20.45-b01, mixed mode)





    2013/4/26 宾莉金 <binlijin@gmail.com>
    Impala download from

    http://beta.cloudera.com/impala/ubuntu/precise/amd64/impala/pool/contrib/i/impala/



    2013/4/26 宾莉金 <binlijin@gmail.com>
    I meet the same problem.
    Ubuntu 12.04.2 LTS




    2013/4/26 Lenni Kuff <lskuff@cloudera.com>
    Hi Rahaul,
    To confirm we have a fix that will work for you, could you please send
    us the hs_err_pid.log file from your crash? From the output you sent,
    this should be located here:

    /var/run/impala/hs_err_pid22887.log

    Thanks,
    Lenni
    Software Engineer - Cloudera

    On Thu, Apr 25, 2013 at 3:30 PM, Miklos Christine <mwc@cloudera.com>
    wrote:
    Hello Rahul,

    I believe this issue has been hit by others on the mailing list.
    See the mailing list for the subject "every impalad daemon crashes".

    There has been a jira filed for the issue and it will be fixed in the
    upcoming GA release, which will be the v1.0.
    Currently, I believe you are on v0.7.1 which still has the issue.
    https://issues.cloudera.org/browse/IMPALA-289

    Thanks,
    Miklos


    On Thu, Apr 25, 2013 at 1:58 PM, Aaron T. Myers <atm@cloudera.com>
    wrote:
    Hi Rahul,

    Moving this over to impala-user@cloudera.org, which should be able
    to help
    you with this issue.


    --
    Aaron T. Myers
    Software Engineer, Cloudera


    On Thu, Apr 25, 2013 at 6:05 AM, rahul mathur
    wrote:
    Hi,

    I am facing an error : "Error communicating with impalad: TSocket
    read 0
    bytes" which causes a JRE crash, this error results when I run
    select query
    on impala-shell. Following are the logs generated after this error in
    impala_server.log :

    13/04/25 11:09:23 INFO service.Frontend: analyze query select * from
    ex
    13/04/25 11:09:23 INFO catalog.HdfsTable: load table ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: load partition block md
    for ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded partition
    PartitionBlockMetadata{#blocks=0, #filenames=0, totalStringLen=0}
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded partition
    PartitionBlockMetadata{#blocks=1, #filenames=1, totalStringLen=104}
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded disk ids for table
    default.ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: 2
    13/04/25 11:09:24 INFO catalog.HdfsTable: table #rows=0
    13/04/25 11:09:24 INFO service.Frontend: create plan
    13/04/25 11:09:24 INFO planner.Planner: create single-node plan
    13/04/25 11:09:24 INFO planner.HdfsScanNode: collecting partitions
    for
    table ex
    13/04/25 11:09:24 INFO planner.HdfsScanNode: finalize HdfsScan:
    cardinality=0
    13/04/25 11:09:24 INFO planner.HdfsScanNode: finalize HdfsScan:
    #nodes=2
    13/04/25 11:09:24 INFO planner.Planner: create plan fragments
    13/04/25 11:09:24 INFO planner.Planner: finalize plan fragments
    13/04/25 11:09:25 INFO service.Frontend: get scan range locations
    13/04/25 11:09:25 INFO service.Frontend: create result set metadata
    13/04/25 11:09:25 INFO service.JniFrontend: Plan Fragment 0
    UNPARTITIONED
    EXCHANGE (1)
    TUPLE IDS: 0

    Plan Fragment 1
    RANDOM
    STREAM DATA SINK
    EXCHANGE ID: 1
    UNPARTITIONED

    SCAN HDFS table=default.ex #partitions=1 size=265B (0)
    TUPLE IDS: 0

    #
    # A fatal error has been detected by the Java Runtime Environment:
    #
    # SIGILL (0x4) at pc=0x00000000008ed857, pid=22887,
    tid=140607123736320
    #
    # JRE version: 7.0_21-b11
    # Java VM: Java HotSpot(TM) 64-Bit Server VM (23.21-b01 mixed mode
    linux-amd64 compressed oops)
    # Problematic frame:
    # C [impalad+0x4ed857] impala::HdfsScanNode::UpdateCounters()+0x97
    #
    # Failed to write core dump. Core dumps have been disabled. To enable
    core dumping, try "ulimit -c unlimited" before starting Java again
    #
    # An error report file with more information is saved as:
    # /var/run/impala/hs_err_pid22887.log
    #
    # If you would like to submit a bug report, please visit:
    # http://bugreport.sun.com/bugreport/crash.jsp


    There is one more log file generated under the home directory of
    impala
    user like "hs_err_pid22887.log" by JRE.
    I tried running same queries on HIVE which showed correct result from
    local and remote client. Also, tried both 0.7 and latest version 1.0
    but
    both are throwing same error. Please reply with a solution ASAP.

    Thanks in advance,
    Rahul Mathur

    --



    --
    *Best Regards,*
    lijin bin


    --
    *Best Regards,*
    lijin bin


    --
    *Best Regards,*
      lijin bin
  • 宾莉金 at Apr 27, 2013 at 1:36 am
    The code not in https://github.com/cloudera/impala?
    I use the code from https://github.com/cloudera/impala.git has the same
    problem.



    2013/4/27 Skye Wanderman-Milne <skye@cloudera.com>
    It looks like you're also hitting
    https://issues.cloudera.org/browse/IMPALA-289, which has been fixed for
    the v1.0 release. Sorry for the inconvenience.

    Skye

    On Fri, Apr 26, 2013 at 6:55 AM, 宾莉金 wrote:

    The core dump is


    warning: Can't read pathname for load map: Input/output error.
    [Thread debugging using libthread_db enabled]
    Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
    Core was generated by
    `/home/tianzhao.blj/cloudera/impala/be/build/debug/service/impalad
    -state_store_'.
    Program terminated with signal 6, Aborted.
    #0 0x00007f35fd6fa425 in raise () from /lib/x86_64-linux-gnu/libc.so.6
    (gdb) bt
    #0 0x00007f35fd6fa425 in raise () from /lib/x86_64-linux-gnu/libc.so.6
    #1 0x00007f35fd6fdb8b in abort () from /lib/x86_64-linux-gnu/libc.so.6
    #2 0x00007f35fca8fa37 in os::abort(bool) () from
    /usr/lib/jvm/java-6-oracle/jre/lib/amd64/server/libjvm.so
    #3 0x00007f35fcbe3f28 in VMError::report_and_die() () from
    /usr/lib/jvm/java-6-oracle/jre/lib/amd64/server/libjvm.so
    #4 0x00007f35fca963f5 in JVM_handle_linux_signal () from
    /usr/lib/jvm/java-6-oracle/jre/lib/amd64/server/libjvm.so
    #5 0x00007f35fca926fe in signalHandler(int, siginfo*, void*) () from
    /usr/lib/jvm/java-6-oracle/jre/lib/amd64/server/libjvm.so
    #6 <signal handler called>
    #7 impala::HdfsScanNode::UpdateCounters (this=0x37e6800) at
    /home/tianzhao.blj/cloudera/impala/be/src/exec/hdfs-scan-node.cc:842
    #8 0x0000000000e5b9cc in impala::HdfsScanNode::GetNext (this=0x37e6800,
    state=0x51cca80, row_batch=0x45f4a00, eos=0x51c8171)
    at
    /home/tianzhao.blj/cloudera/impala/be/src/exec/hdfs-scan-node.cc:162
    #9 0x0000000000d0af2d in impala::PlanFragmentExecutor::GetNextInternal
    (this=0x51c8038, batch=0x7f35de6203b8)
    at
    /home/tianzhao.blj/cloudera/impala/be/src/runtime/plan-fragment-executor.cc:377
    #10 0x0000000000d09b0a in impala::PlanFragmentExecutor::OpenInternal
    (this=0x51c8038)
    at
    /home/tianzhao.blj/cloudera/impala/be/src/runtime/plan-fragment-executor.cc:249
    #11 0x0000000000d09885 in impala::PlanFragmentExecutor::Open
    (this=0x51c8038)
    at
    /home/tianzhao.blj/cloudera/impala/be/src/runtime/plan-fragment-executor.cc:226
    #12 0x0000000000b323a3 in impala::ImpalaServer::FragmentExecState::Exec
    (this=0x51c8000)
    at
    /home/tianzhao.blj/cloudera/impala/be/src/service/impala-server.cc:470
    #13 0x0000000000b45c67 in impala::ImpalaServer::RunExecPlanFragment
    (this=0x3cddc00, exec_state=0x51c8000)
    at
    /home/tianzhao.blj/cloudera/impala/be/src/service/impala-server.cc:1692
    #14 0x0000000000bad05f in boost::_mfi::mf1<void, impala::ImpalaServer,
    impala::ImpalaServer::FragmentExecState*>::operator() (
    this=0x2b1f308, p=0x3cddc00, a1=0x51c8000) at
    /usr/include/boost/bind/mem_fn_template.hpp:165
    #15 0x0000000000bac741 in
    boost::_bi::list2<boost::_bi::value<impala::ImpalaServer*>,
    boost::_bi::value<impala::ImpalaServer::FragmentExecState*>
    ::operator()<boost::_mfi::mf1<void, impala::ImpalaServer,
    impala::ImpalaServer::FragmentExecState*>, boost::_bi::list0>
    (this=0x2b1f318, f=..., a=...) at /usr/include/boost/bind/bind.hpp:313
    #16 0x0000000000bab717 in boost::_bi::bind_t<void, boost::_mfi::mf1<void,
    impala::ImpalaServer, impala::ImpalaServer::FragmentExecState*>,
    boost::_bi::list2<boost::_bi::value<impala::ImpalaServer*>,
    boost::_bi::value<impala::ImpalaServer::FragmentExecState*> > >::operator()
    (this=0x2b1f308) at /usr/include/boost/bind/bind_template.hpp:20
    #17 0x0000000000baa48a in
    boost::detail::thread_data<boost::_bi::bind_t<void, boost::_mfi::mf1<void,
    impala::ImpalaServer, impala::ImpalaServer::FragmentExecState*>,
    boost::_bi::list2<boost::_bi::value<impala::ImpalaServer*>,
    boost::_bi::value<impala::ImpalaServer::FragmentExecState*> > > >::run
    (this=0x2b1f180) at /usr/include/boost/thread/detail/thread.hpp:61
    #18 0x00007f35ff62ace9 in thread_proxy () from
    /usr/lib/libboost_thread.so.1.46.1
    #19 0x00007f35ff408e9a in start_thread () from
    /lib/x86_64-linux-gnu/libpthread.so.0
    #20 0x00007f35fd7b7ccd in clone () from /lib/x86_64-linux-gnu/libc.so.6
    #21 0x0000000000000000 in ?? ()



    java version "1.6.0_45"
    Java(TM) SE Runtime Environment (build 1.6.0_45-b06)
    Java HotSpot(TM) 64-Bit Server VM (build 20.45-b01, mixed mode)





    2013/4/26 宾莉金 <binlijin@gmail.com>
    Impala download from

    http://beta.cloudera.com/impala/ubuntu/precise/amd64/impala/pool/contrib/i/impala/



    2013/4/26 宾莉金 <binlijin@gmail.com>
    I meet the same problem.
    Ubuntu 12.04.2 LTS




    2013/4/26 Lenni Kuff <lskuff@cloudera.com>
    Hi Rahaul,
    To confirm we have a fix that will work for you, could you please send
    us the hs_err_pid.log file from your crash? From the output you sent,
    this should be located here:

    /var/run/impala/hs_err_pid22887.log

    Thanks,
    Lenni
    Software Engineer - Cloudera

    On Thu, Apr 25, 2013 at 3:30 PM, Miklos Christine <mwc@cloudera.com>
    wrote:
    Hello Rahul,

    I believe this issue has been hit by others on the mailing list.
    See the mailing list for the subject "every impalad daemon crashes".

    There has been a jira filed for the issue and it will be fixed in the
    upcoming GA release, which will be the v1.0.
    Currently, I believe you are on v0.7.1 which still has the issue.
    https://issues.cloudera.org/browse/IMPALA-289

    Thanks,
    Miklos


    On Thu, Apr 25, 2013 at 1:58 PM, Aaron T. Myers <atm@cloudera.com>
    wrote:
    Hi Rahul,

    Moving this over to impala-user@cloudera.org, which should be able
    to help
    you with this issue.


    --
    Aaron T. Myers
    Software Engineer, Cloudera


    On Thu, Apr 25, 2013 at 6:05 AM, rahul mathur
    wrote:
    Hi,

    I am facing an error : "Error communicating with impalad: TSocket
    read 0
    bytes" which causes a JRE crash, this error results when I run
    select query
    on impala-shell. Following are the logs generated after this error
    in
    impala_server.log :

    13/04/25 11:09:23 INFO service.Frontend: analyze query select *
    from ex
    13/04/25 11:09:23 INFO catalog.HdfsTable: load table ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: load partition block md
    for ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded partition
    PartitionBlockMetadata{#blocks=0, #filenames=0, totalStringLen=0}
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded partition
    PartitionBlockMetadata{#blocks=1, #filenames=1, totalStringLen=104}
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded disk ids for table
    default.ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: 2
    13/04/25 11:09:24 INFO catalog.HdfsTable: table #rows=0
    13/04/25 11:09:24 INFO service.Frontend: create plan
    13/04/25 11:09:24 INFO planner.Planner: create single-node plan
    13/04/25 11:09:24 INFO planner.HdfsScanNode: collecting partitions
    for
    table ex
    13/04/25 11:09:24 INFO planner.HdfsScanNode: finalize HdfsScan:
    cardinality=0
    13/04/25 11:09:24 INFO planner.HdfsScanNode: finalize HdfsScan:
    #nodes=2
    13/04/25 11:09:24 INFO planner.Planner: create plan fragments
    13/04/25 11:09:24 INFO planner.Planner: finalize plan fragments
    13/04/25 11:09:25 INFO service.Frontend: get scan range locations
    13/04/25 11:09:25 INFO service.Frontend: create result set metadata
    13/04/25 11:09:25 INFO service.JniFrontend: Plan Fragment 0
    UNPARTITIONED
    EXCHANGE (1)
    TUPLE IDS: 0

    Plan Fragment 1
    RANDOM
    STREAM DATA SINK
    EXCHANGE ID: 1
    UNPARTITIONED

    SCAN HDFS table=default.ex #partitions=1 size=265B (0)
    TUPLE IDS: 0

    #
    # A fatal error has been detected by the Java Runtime Environment:
    #
    # SIGILL (0x4) at pc=0x00000000008ed857, pid=22887,
    tid=140607123736320
    #
    # JRE version: 7.0_21-b11
    # Java VM: Java HotSpot(TM) 64-Bit Server VM (23.21-b01 mixed mode
    linux-amd64 compressed oops)
    # Problematic frame:
    # C [impalad+0x4ed857]
    impala::HdfsScanNode::UpdateCounters()+0x97
    #
    # Failed to write core dump. Core dumps have been disabled. To
    enable
    core dumping, try "ulimit -c unlimited" before starting Java again
    #
    # An error report file with more information is saved as:
    # /var/run/impala/hs_err_pid22887.log
    #
    # If you would like to submit a bug report, please visit:
    # http://bugreport.sun.com/bugreport/crash.jsp


    There is one more log file generated under the home directory of
    impala
    user like "hs_err_pid22887.log" by JRE.
    I tried running same queries on HIVE which showed correct result
    from
    local and remote client. Also, tried both 0.7 and latest version
    1.0 but
    both are throwing same error. Please reply with a solution ASAP.

    Thanks in advance,
    Rahul Mathur

    --



    --
    *Best Regards,*
    lijin bin


    --
    *Best Regards,*
    lijin bin


    --
    *Best Regards,*
    lijin bin

    --
    *Best Regards,*
      lijin bin
  • Rahul mathur at Apr 30, 2013 at 5:53 am
    @Nong Li : thanx for sharing the issue, that is really helpful to know what
    was causing that error.
    @Lenni : Its hard for me to look into the C code as it is not my forte,
    Java is. Only option left with me is to wait for new release :).

    Thanks,
    *Rahul Mathur*

    On Tue, Apr 30, 2013 at 5:13 AM, Lenni Kuff wrote:

    This fix has been pushed to the Impala githup repo if you are interested
    in digging into the code:


    https://github.com/cloudera/impala/commit/07dc2a206bc762fe31591a7748b180604787baca

    Thanks,
    Lenni

    On Mon, Apr 29, 2013 at 9:52 AM, Nong Li wrote:

    The issue is that we were using a hardware instruction (popcnt) without
    first checking if
    the chip supported it. The fix is simply to fall back to a less
    optimized implementation
    on older (pre-nehalem for intel, pre-barcelona for amd) machines.

    The 1.0 release is coming out soon and will contain this fix.


    On Mon, Apr 29, 2013 at 3:01 AM, rahul mathur <
    rahulmathur.email@gmail.com> wrote:
    Hi lskuff,
    *
    *
    I was not in touch with my mails from past few days, Whereas I can see
    same logs as posted by* *宾莉金 at my end. I followed the mail trail and
    so is the solution suggested impala *V1.0*.
    I am waiting for its release impatiently now and hope to download it
    soon. Also, if any one could explain what the problem was and fix is,
    that will really be appreciated.

    Thanks in advance,
    Rahul Mathur
    *
    *
    On Friday, April 26, 2013 2:28:09 AM UTC+5:30, Aaron T. Myers wrote:

    Hi Rahul,

    Moving this over to impal...@cloudera.org, which should be able to
    help you with this issue.


    --
    Aaron T. Myers
    Software Engineer, Cloudera

    On Thu, Apr 25, 2013 at 6:05 AM, rahul mathur wrote:

    Hi,

    I am facing an error : "*Error communicating with impalad: TSocket
    read 0 bytes" *which causes a JRE crash, this error results when I
    run select query on impala-shell. Following are the logs generated after
    this error in impala_server.log :

    13/04/25 11:09:23 INFO service.Frontend: analyze query select * from ex
    13/04/25 11:09:23 INFO catalog.HdfsTable: load table ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: load partition block md for
    ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded partition
    PartitionBlockMetadata{#**blocks=0, #filenames=0, totalStringLen=0}
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded partition
    PartitionBlockMetadata{#**blocks=1, #filenames=1, totalStringLen=104}
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded disk ids for table
    default.ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: 2
    13/04/25 11:09:24 INFO catalog.HdfsTable: table #rows=0
    13/04/25 11:09:24 INFO service.Frontend: create plan
    13/04/25 11:09:24 INFO planner.Planner: create single-node plan
    13/04/25 11:09:24 INFO planner.HdfsScanNode: collecting partitions for
    table ex
    13/04/25 11:09:24 INFO planner.HdfsScanNode: finalize HdfsScan:
    cardinality=0
    13/04/25 11:09:24 INFO planner.HdfsScanNode: finalize HdfsScan:
    #nodes=2
    13/04/25 11:09:24 INFO planner.Planner: create plan fragments
    13/04/25 11:09:24 INFO planner.Planner: finalize plan fragments
    13/04/25 11:09:25 INFO service.Frontend: get scan range locations
    13/04/25 11:09:25 INFO service.Frontend: create result set metadata
    13/04/25 11:09:25 INFO service.JniFrontend: Plan Fragment 0
    UNPARTITIONED
    EXCHANGE (1)
    TUPLE IDS: 0

    Plan Fragment 1
    RANDOM
    STREAM DATA SINK
    EXCHANGE ID: 1
    UNPARTITIONED

    SCAN HDFS table=default.ex #partitions=1 size=265B (0)
    TUPLE IDS: 0

    #
    # A fatal error has been detected by the Java Runtime Environment:
    #
    # SIGILL (0x4) at pc=0x00000000008ed857, pid=22887,
    tid=140607123736320
    #
    # JRE version: 7.0_21-b11
    # Java VM: Java HotSpot(TM) 64-Bit Server VM (23.21-b01 mixed mode
    linux-amd64 compressed oops)
    # Problematic frame:
    # C [impalad+0x4ed857] * impala::HdfsScanNode::UpdateCounters()+0x97*
    #
    # Failed to write core dump. Core dumps have been disabled. To enable
    core dumping, try "ulimit -c unlimited" before starting Java again
    #
    # An error report file with more information is saved as:
    # */var/run/impala/hs_err_pid22887.log*
    #
    # If you would like to submit a bug report, please visit:
    # http://bugreport.sun.com/**bugreport/crash.jsp<http://bugreport.sun.com/bugreport/crash.jsp>


    There is one more log file generated under the home directory of
    impala user like "hs_err_pid22887.log" by JRE.
    I tried running same queries on HIVE which showed correct result from
    local and remote client. Also, tried both 0.7 and latest version 1.0 but
    both are throwing same error. Please reply with a solution ASAP.

    Thanks in advance,
    *Rahul Mathur*

    --


  • Amigo at May 30, 2013 at 4:11 am
    So, started a brand new installed using CDH 4.5.2

    *Version*: 4.5.2 (#327 built by jenkins on 20130429-1453 git:
    16cab2c7b76194b7877d64a4215494daa387a266)
    impala-shell -v
    Impala Shell v1.0 (d1bf0d1) built on Sun Apr 28 15:33:52 PDT 2013

    Using IMPALA 1.0-1.p0.371 parcel

        - MR jobs work
        - HiveQL works (for same queries stated below)
        - Impala
           - Simple select works (in hive too)
           - Queries with group fails

    Query: select search_word, count(*) from cs group by search_word

    Error communicating with impalad: TSocket read 0 bytes



    Anybody spot what I may be doing wrong? Missing any config? I couldn't find
    anything useful in /var/logs/impala/impalad.INFO
    What else can I provide?

    thanks, nitin

    On Thursday, May 2, 2013 5:42:06 AM UTC-5, rahul mathur wrote:

    Hi All,

    As suggested impala v1.0 has solved that issue I was facing. Thanks to all
    Cloudera's engineers for making it a hit.

    Regards,
    *Rahul Mathur*


    On Tue, Apr 30, 2013 at 11:23 AM, rahul mathur <rahulmat...@gmail.com<javascript:>
    wrote:
    @Nong Li : thanx for sharing the issue, that is really helpful to know
    what was causing that error.
    @Lenni : Its hard for me to look into the C code as it is not my forte,
    Java is. Only option left with me is to wait for new release :).

    Thanks,
    *Rahul Mathur*


    On Tue, Apr 30, 2013 at 5:13 AM, Lenni Kuff <lsk...@cloudera.com<javascript:>
    wrote:
    This fix has been pushed to the Impala githup repo if you are interested
    in digging into the code:


    https://github.com/cloudera/impala/commit/07dc2a206bc762fe31591a7748b180604787baca

    Thanks,
    Lenni


    On Mon, Apr 29, 2013 at 9:52 AM, Nong Li <no...@cloudera.com<javascript:>
    wrote:
    The issue is that we were using a hardware instruction (popcnt) without
    first checking if
    the chip supported it. The fix is simply to fall back to a less
    optimized implementation
    on older (pre-nehalem for intel, pre-barcelona for amd) machines.

    The 1.0 release is coming out soon and will contain this fix.


    On Mon, Apr 29, 2013 at 3:01 AM, rahul mathur <rahulmat...@gmail.com<javascript:>
    wrote:
    Hi lskuff,
    *
    *
    I was not in touch with my mails from past few days, Whereas I can see
    same logs as posted by* *宾莉金 at my end. I followed the mail trail and
    so is the solution suggested impala *V1.0*.
    I am waiting for its release impatiently now and hope to download it
    soon. Also, if any one could explain what the problem was and fix is,
    that will really be appreciated.

    Thanks in advance,
    Rahul Mathur
    *
    *
    On Friday, April 26, 2013 2:28:09 AM UTC+5:30, Aaron T. Myers wrote:

    Hi Rahul,

    Moving this over to impal...@cloudera.org, which should be able to
    help you with this issue.


    --
    Aaron T. Myers
    Software Engineer, Cloudera

    On Thu, Apr 25, 2013 at 6:05 AM, rahul mathur wrote:

    Hi,

    I am facing an error : "*Error communicating with impalad: TSocket
    read 0 bytes" *which causes a JRE crash, this error results when I
    run select query on impala-shell. Following are the logs generated after
    this error in impala_server.log :

    13/04/25 11:09:23 INFO service.Frontend: analyze query select * from
    ex
    13/04/25 11:09:23 INFO catalog.HdfsTable: load table ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: load partition block md
    for ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded partition
    PartitionBlockMetadata{#**blocks=0, #filenames=0, totalStringLen=0}
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded partition
    PartitionBlockMetadata{#**blocks=1, #filenames=1,
    totalStringLen=104}
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded disk ids for table
    default.ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: 2
    13/04/25 11:09:24 INFO catalog.HdfsTable: table #rows=0
    13/04/25 11:09:24 INFO service.Frontend: create plan
    13/04/25 11:09:24 INFO planner.Planner: create single-node plan
    13/04/25 11:09:24 INFO planner.HdfsScanNode: collecting partitions
    for table ex
    13/04/25 11:09:24 INFO planner.HdfsScanNode: finalize HdfsScan:
    cardinality=0
    13/04/25 11:09:24 INFO planner.HdfsScanNode: finalize HdfsScan:
    #nodes=2
    13/04/25 11:09:24 INFO planner.Planner: create plan fragments
    13/04/25 11:09:24 INFO planner.Planner: finalize plan fragments
    13/04/25 11:09:25 INFO service.Frontend: get scan range locations
    13/04/25 11:09:25 INFO service.Frontend: create result set metadata
    13/04/25 11:09:25 INFO service.JniFrontend: Plan Fragment 0
    UNPARTITIONED
    EXCHANGE (1)
    TUPLE IDS: 0

    Plan Fragment 1
    RANDOM
    STREAM DATA SINK
    EXCHANGE ID: 1
    UNPARTITIONED

    SCAN HDFS table=default.ex #partitions=1 size=265B (0)
    TUPLE IDS: 0

    #
    # A fatal error has been detected by the Java Runtime Environment:
    #
    # SIGILL (0x4) at pc=0x00000000008ed857, pid=22887,
    tid=140607123736320
    #
    # JRE version: 7.0_21-b11
    # Java VM: Java HotSpot(TM) 64-Bit Server VM (23.21-b01 mixed mode
    linux-amd64 compressed oops)
    # Problematic frame:
    # C [impalad+0x4ed857] * impala::HdfsScanNode::
    UpdateCounters()+0x97*
    #
    # Failed to write core dump. Core dumps have been disabled. To
    enable core dumping, try "ulimit -c unlimited" before starting Java again
    #
    # An error report file with more information is saved as:
    # */var/run/impala/hs_err_pid22887.log*
    #
    # If you would like to submit a bug report, please visit:
    # http://bugreport.sun.com/**bugreport/crash.jsp<http://bugreport.sun.com/bugreport/crash.jsp>


    There is one more log file generated under the home directory of
    impala user like "hs_err_pid22887.log" by JRE.
    I tried running same queries on HIVE which showed correct result
    from local and remote client. Also, tried both 0.7 and latest version 1.0
    but both are throwing same error. Please reply with a solution ASAP.

    Thanks in advance,
    *Rahul Mathur*

    --


  • Lenni Kuff at May 30, 2013 at 3:19 pm
    Hi Nitin,
    Could you please provide the impalad logs so we can help investigate this
    problem? It would also be useful to have "describe extended" output of your
    target table. You can get this by running:
    hive -e "describe extended cs"
    Thanks,
    Lenni
    Software Engineer - Cloudera
    On Wed, May 29, 2013 at 9:11 PM, Amigo wrote:

    So, started a brand new installed using CDH 4.5.2

    *Version*: 4.5.2 (#327 built by jenkins on 20130429-1453 git:
    16cab2c7b76194b7877d64a4215494daa387a266)
    impala-shell -v
    Impala Shell v1.0 (d1bf0d1) built on Sun Apr 28 15:33:52 PDT 2013

    Using IMPALA 1.0-1.p0.371 parcel

    - MR jobs work
    - HiveQL works (for same queries stated below)
    - Impala
    - Simple select works (in hive too)
    - Queries with group fails

    Query: select search_word, count(*) from cs group by search_word

    Error communicating with impalad: TSocket read 0 bytes



    Anybody spot what I may be doing wrong? Missing any config? I couldn't
    find anything useful in /var/logs/impala/impalad.INFO
    What else can I provide?

    thanks, nitin

    On Thursday, May 2, 2013 5:42:06 AM UTC-5, rahul mathur wrote:

    Hi All,

    As suggested impala v1.0 has solved that issue I was facing. Thanks to
    all Cloudera's engineers for making it a hit.

    Regards,
    *Rahul Mathur*

    On Tue, Apr 30, 2013 at 11:23 AM, rahul mathur wrote:

    @Nong Li : thanx for sharing the issue, that is really helpful to know
    what was causing that error.
    @Lenni : Its hard for me to look into the C code as it is not my forte,
    Java is. Only option left with me is to wait for new release :).

    Thanks,
    *Rahul Mathur*

    On Tue, Apr 30, 2013 at 5:13 AM, Lenni Kuff wrote:

    This fix has been pushed to the Impala githup repo if you are
    interested in digging into the code:

    https://github.com/cloudera/**impala/commit/**
    07dc2a206bc762fe31591a7748b180**604787baca<https://github.com/cloudera/impala/commit/07dc2a206bc762fe31591a7748b180604787baca>

    Thanks,
    Lenni

    On Mon, Apr 29, 2013 at 9:52 AM, Nong Li wrote:

    The issue is that we were using a hardware instruction (popcnt)
    without first checking if
    the chip supported it. The fix is simply to fall back to a less
    optimized implementation
    on older (pre-nehalem for intel, pre-barcelona for amd) machines.

    The 1.0 release is coming out soon and will contain this fix.

    On Mon, Apr 29, 2013 at 3:01 AM, rahul mathur wrote:


    Hi lskuff,
    *
    *
    I was not in touch with my mails from past few days, Whereas I can
    see same logs as posted by* *宾莉金 at my end. I followed the mail
    trail and so is the solution suggested impala *V1.0*.
    I am waiting for its release impatiently now and hope to download it
    soon. Also, if any one could explain what the problem was and fix
    is, that will really be appreciated.

    Thanks in advance,
    Rahul Mathur
    *
    *
    On Friday, April 26, 2013 2:28:09 AM UTC+5:30, Aaron T. Myers wrote:

    Hi Rahul,

    Moving this over to impal...@cloudera.org, which should be able to
    help you with this issue.


    --
    Aaron T. Myers
    Software Engineer, Cloudera


    On Thu, Apr 25, 2013 at 6:05 AM, rahul mathur <rahulmat...@gmail.com
    wrote:
    Hi,

    I am facing an error : "*Error communicating with impalad: TSocket
    read 0 bytes" *which causes a JRE crash, this error results when I
    run select query on impala-shell. Following are the logs generated after
    this error in impala_server.log :

    13/04/25 11:09:23 INFO service.Frontend: analyze query select *
    from ex
    13/04/25 11:09:23 INFO catalog.HdfsTable: load table ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: load partition block md
    for ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded partition
    PartitionBlockMetadata{#**blocks**=0, #filenames=0,
    totalStringLen=0}
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded partition
    PartitionBlockMetadata{#**blocks**=1, #filenames=1,
    totalStringLen=104}
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded disk ids for table
    default.ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: 2
    13/04/25 11:09:24 INFO catalog.HdfsTable: table #rows=0
    13/04/25 11:09:24 INFO service.Frontend: create plan
    13/04/25 11:09:24 INFO planner.Planner: create single-node plan
    13/04/25 11:09:24 INFO planner.HdfsScanNode: collecting partitions
    for table ex
    13/04/25 11:09:24 INFO planner.HdfsScanNode: finalize HdfsScan:
    cardinality=0
    13/04/25 11:09:24 INFO planner.HdfsScanNode: finalize HdfsScan:
    #nodes=2
    13/04/25 11:09:24 INFO planner.Planner: create plan fragments
    13/04/25 11:09:24 INFO planner.Planner: finalize plan fragments
    13/04/25 11:09:25 INFO service.Frontend: get scan range locations
    13/04/25 11:09:25 INFO service.Frontend: create result set metadata
    13/04/25 11:09:25 INFO service.JniFrontend: Plan Fragment 0
    UNPARTITIONED
    EXCHANGE (1)
    TUPLE IDS: 0

    Plan Fragment 1
    RANDOM
    STREAM DATA SINK
    EXCHANGE ID: 1
    UNPARTITIONED

    SCAN HDFS table=default.ex #partitions=1 size=265B (0)
    TUPLE IDS: 0

    #
    # A fatal error has been detected by the Java Runtime Environment:
    #
    # SIGILL (0x4) at pc=0x00000000008ed857, pid=22887,
    tid=140607123736320
    #
    # JRE version: 7.0_21-b11
    # Java VM: Java HotSpot(TM) 64-Bit Server VM (23.21-b01 mixed mode
    linux-amd64 compressed oops)
    # Problematic frame:
    # C [impalad+0x4ed857] * impala::HdfsScanNode::UpdateC
    ounters()+0x97*
    #
    # Failed to write core dump. Core dumps have been disabled. To
    enable core dumping, try "ulimit -c unlimited" before starting Java again
    #
    # An error report file with more information is saved as:
    # */var/run/impala/hs_err_pid22887.log*
    #
    # If you would like to submit a bug report, please visit:
    # http://bugreport.sun.com/**bugre**port/crash.jsp<http://bugreport.sun.com/bugreport/crash.jsp>


    There is one more log file generated under the home directory of
    impala user like "hs_err_pid22887.log" by JRE.
    I tried running same queries on HIVE which showed correct result
    from local and remote client. Also, tried both 0.7 and latest version 1.0
    but both are throwing same error. Please reply with a solution ASAP.

    Thanks in advance,
    *Rahul Mathur*

    --


  • Amigo at May 30, 2013 at 4:08 pm
    Hive output

    Logging initialized using configuration in
    jar:file:/opt/cloudera/parcels/CDH-4.2.1-1.cdh4.2.1.p0.5/lib/hive/lib/hive-common-0.10.0-cdh4.2.1.jar!/hive-log4j.properties
    Hive history file=/tmp/ME/hive_job_log_ME_201305301035_1001543262.txt

    OK

    first_visit_date string

    acquisition_source_url string

    first_url string

    ip_address string

    browser_os_name string



    Detailed Table Information Table(tableName:cs, dbName:mobile, owner:ME,
    createTime:1369863897, lastAccessTime:0, retention:0,
    sd:StorageDescriptor(cols:[FieldSchema(name:first_visit_date, type:string,
    comment:null), FieldSchema(name:acquisition_source_url, type:string,
    comment:null), FieldSchema(name:first_url, type:string, comment:null),
    FieldSchema(name:ip_address, type:string, comment:null),
    FieldSchema(name:browser_os_name, type:string, comment:null)],
    location:hdfs://foobar-1.com:8020/bi/mobile/cs,
    inputFormat:org.apache.hadoop.mapred.TextInputFormat,
    outputFormat:org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat,
    compressed:false, numBuckets:0, serdeInfo:SerDeInfo(name:null,
    serializationLib:org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe,
    parameters:{serialization.format= , field.delim=
    Table creation:
    drop table if exists cs;

    create external table cs

    (

    FIRST_VISIT_DATE string,ACQUISITION_SOURCE_URL string,DESTINATION_URL
    string,IP_ADDRESS string,BROWSER_OS_NAME string
    )

    row format delimited fields terminated by '\t'

    location '/mobile/cs';
    Logs I found in /var/logs/impalad are here - http://pastebin.com/BgtnRq7Y

    thanks, nitin

    On Thursday, May 30, 2013 10:19:41 AM UTC-5, lskuff wrote:

    Hi Nitin,
    Could you please provide the impalad logs so we can help investigate this
    problem? It would also be useful to have "describe extended" output of your
    target table. You can get this by running:
    hive -e "describe extended cs"
    Thanks,
    Lenni
    Software Engineer - Cloudera

    On Wed, May 29, 2013 at 9:11 PM, Amigo <nitin....@gmail.com <javascript:>>wrote:
    So, started a brand new installed using CDH 4.5.2

    *Version*: 4.5.2 (#327 built by jenkins on 20130429-1453 git:
    16cab2c7b76194b7877d64a4215494daa387a266)
    impala-shell -v
    Impala Shell v1.0 (d1bf0d1) built on Sun Apr 28 15:33:52 PDT 2013

    Using IMPALA 1.0-1.p0.371 parcel

    - MR jobs work
    - HiveQL works (for same queries stated below)
    - Impala
    - Simple select works (in hive too)
    - Queries with group fails

    Query: select search_word, count(*) from cs group by search_word

    Error communicating with impalad: TSocket read 0 bytes



    Anybody spot what I may be doing wrong? Missing any config? I couldn't
    find anything useful in /var/logs/impala/impalad.INFO
    What else can I provide?

    thanks, nitin

    On Thursday, May 2, 2013 5:42:06 AM UTC-5, rahul mathur wrote:

    Hi All,

    As suggested impala v1.0 has solved that issue I was facing. Thanks to
    all Cloudera's engineers for making it a hit.

    Regards,
    *Rahul Mathur*

    On Tue, Apr 30, 2013 at 11:23 AM, rahul mathur wrote:

    @Nong Li : thanx for sharing the issue, that is really helpful to
    know what was causing that error.
    @Lenni : Its hard for me to look into the C code as it is not my
    forte, Java is. Only option left with me is to wait for new release :).

    Thanks,
    *Rahul Mathur*

    On Tue, Apr 30, 2013 at 5:13 AM, Lenni Kuff wrote:

    This fix has been pushed to the Impala githup repo if you are
    interested in digging into the code:

    https://github.com/cloudera/**impala/commit/**
    07dc2a206bc762fe31591a7748b180**604787baca<https://github.com/cloudera/impala/commit/07dc2a206bc762fe31591a7748b180604787baca>

    Thanks,
    Lenni

    On Mon, Apr 29, 2013 at 9:52 AM, Nong Li wrote:

    The issue is that we were using a hardware instruction (popcnt)
    without first checking if
    the chip supported it. The fix is simply to fall back to a less
    optimized implementation
    on older (pre-nehalem for intel, pre-barcelona for amd) machines.

    The 1.0 release is coming out soon and will contain this fix.

    On Mon, Apr 29, 2013 at 3:01 AM, rahul mathur wrote:


    Hi lskuff,
    *
    *
    I was not in touch with my mails from past few days, Whereas I can
    see same logs as posted by* *宾莉金 at my end. I followed the mail
    trail and so is the solution suggested impala *V1.0*.
    I am waiting for its release impatiently now and hope to download it
    soon. Also, if any one could explain what the problem was and fix
    is, that will really be appreciated.

    Thanks in advance,
    Rahul Mathur
    *
    *
    On Friday, April 26, 2013 2:28:09 AM UTC+5:30, Aaron T. Myers wrote:

    Hi Rahul,

    Moving this over to impal...@cloudera.org, which should be able to
    help you with this issue.


    --
    Aaron T. Myers
    Software Engineer, Cloudera


    On Thu, Apr 25, 2013 at 6:05 AM, rahul mathur <
    rahulmat...@gmail.com> wrote:
    Hi,

    I am facing an error : "*Error communicating with impalad:
    TSocket read 0 bytes" *which causes a JRE crash, this error
    results when I run select query on impala-shell. Following are the logs
    generated after this error in impala_server.log :

    13/04/25 11:09:23 INFO service.Frontend: analyze query select *
    from ex
    13/04/25 11:09:23 INFO catalog.HdfsTable: load table ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: load partition block md
    for ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded partition
    PartitionBlockMetadata{#**blocks**=0, #filenames=0,
    totalStringLen=0}
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded partition
    PartitionBlockMetadata{#**blocks**=1, #filenames=1,
    totalStringLen=104}
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded disk ids for
    table default.ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: 2
    13/04/25 11:09:24 INFO catalog.HdfsTable: table #rows=0
    13/04/25 11:09:24 INFO service.Frontend: create plan
    13/04/25 11:09:24 INFO planner.Planner: create single-node plan
    13/04/25 11:09:24 INFO planner.HdfsScanNode: collecting partitions
    for table ex
    13/04/25 11:09:24 INFO planner.HdfsScanNode: finalize HdfsScan:
    cardinality=0
    13/04/25 11:09:24 INFO planner.HdfsScanNode: finalize HdfsScan:
    #nodes=2
    13/04/25 11:09:24 INFO planner.Planner: create plan fragments
    13/04/25 11:09:24 INFO planner.Planner: finalize plan fragments
    13/04/25 11:09:25 INFO service.Frontend: get scan range locations
    13/04/25 11:09:25 INFO service.Frontend: create result set metadata
    13/04/25 11:09:25 INFO service.JniFrontend: Plan Fragment 0
    UNPARTITIONED
    EXCHANGE (1)
    TUPLE IDS: 0

    Plan Fragment 1
    RANDOM
    STREAM DATA SINK
    EXCHANGE ID: 1
    UNPARTITIONED

    SCAN HDFS table=default.ex #partitions=1 size=265B (0)
    TUPLE IDS: 0

    #
    # A fatal error has been detected by the Java Runtime Environment:
    #
    # SIGILL (0x4) at pc=0x00000000008ed857, pid=22887,
    tid=140607123736320
    #
    # JRE version: 7.0_21-b11
    # Java VM: Java HotSpot(TM) 64-Bit Server VM (23.21-b01 mixed mode
    linux-amd64 compressed oops)
    # Problematic frame:
    # C [impalad+0x4ed857] * impala::HdfsScanNode::UpdateC
    ounters()+0x97*
    #
    # Failed to write core dump. Core dumps have been disabled. To
    enable core dumping, try "ulimit -c unlimited" before starting Java again
    #
    # An error report file with more information is saved as:
    # */var/run/impala/hs_err_pid22887.log*
    #
    # If you would like to submit a bug report, please visit:
    # http://bugreport.sun.com/**bugre**port/crash.jsp<http://bugreport.sun.com/bugreport/crash.jsp>


    There is one more log file generated under the home directory of
    impala user like "hs_err_pid22887.log" by JRE.
    I tried running same queries on HIVE which showed correct result
    from local and remote client. Also, tried both 0.7 and latest version 1.0
    but both are throwing same error. Please reply with a solution ASAP.

    Thanks in advance,
    *Rahul Mathur*

    --


  • Lenni Kuff at Jun 1, 2013 at 3:03 pm
    Thanks for the information. I don't see anything immediately obvious from
    the logs of the table definition. In your browser could you go to:
    http://<hostname>:25000 on one of the machines running Impalad? You will
    see a "Hardware Info" box, could you send that output back?

    Thanks,
    Lenni
    On Thu, May 30, 2013 at 9:08 AM, Amigo wrote:

    Hive output

    Logging initialized using configuration in
    jar:file:/opt/cloudera/parcels/CDH-4.2.1-1.cdh4.2.1.p0.5/lib/hive/lib/hive-common-0.10.0-cdh4.2.1.jar!/hive-log4j.properties
    Hive history file=/tmp/ME/hive_job_log_ME_201305301035_1001543262.txt

    OK

    first_visit_date string

    acquisition_source_url string

    first_url string

    ip_address string

    browser_os_name string



    Detailed Table Information Table(tableName:cs, dbName:mobile, owner:ME,
    createTime:1369863897, lastAccessTime:0, retention:0,
    sd:StorageDescriptor(cols:[FieldSchema(name:first_visit_date, type:string,
    comment:null), FieldSchema(name:acquisition_source_url, type:string,
    comment:null), FieldSchema(name:first_url, type:string, comment:null),
    FieldSchema(name:ip_address, type:string, comment:null),
    FieldSchema(name:browser_os_name, type:string, comment:null)],
    location:hdfs://foobar-1.com:8020/bi/mobile/cs,
    inputFormat:org.apache.hadoop.mapred.TextInputFormat,
    outputFormat:org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat,
    compressed:false, numBuckets:0, serdeInfo:SerDeInfo(name:null,
    serializationLib:org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe,
    parameters:{serialization.format= , field.delim=
    Table creation:
    drop table if exists cs;

    create external table cs

    (

    FIRST_VISIT_DATE string,ACQUISITION_SOURCE_URL string,DESTINATION_URL
    string,IP_ADDRESS string,BROWSER_OS_NAME string
    )

    row format delimited fields terminated by '\t'

    location '/mobile/cs';
    Logs I found in /var/logs/impalad are here - http://pastebin.com/BgtnRq7Y

    thanks, nitin

    On Thursday, May 30, 2013 10:19:41 AM UTC-5, lskuff wrote:

    Hi Nitin,
    Could you please provide the impalad logs so we can help investigate this
    problem? It would also be useful to have "describe extended" output of your
    target table. You can get this by running:
    hive -e "describe extended cs"
    Thanks,
    Lenni
    Software Engineer - Cloudera

    On Wed, May 29, 2013 at 9:11 PM, Amigo wrote:

    So, started a brand new installed using CDH 4.5.2

    *Version*: 4.5.2 (#327 built by jenkins on 20130429-1453 git:
    16cab2c7b76194b7877d64a4215494**daa387a266)
    impala-shell -v
    Impala Shell v1.0 (d1bf0d1) built on Sun Apr 28 15:33:52 PDT 2013

    Using IMPALA 1.0-1.p0.371 parcel

    - MR jobs work
    - HiveQL works (for same queries stated below)
    - Impala
    - Simple select works (in hive too)
    - Queries with group fails

    Query: select search_word, count(*) from cs group by search_word

    Error communicating with impalad: TSocket read 0 bytes



    Anybody spot what I may be doing wrong? Missing any config? I couldn't
    find anything useful in /var/logs/impala/impalad.INFO
    What else can I provide?

    thanks, nitin

    On Thursday, May 2, 2013 5:42:06 AM UTC-5, rahul mathur wrote:

    Hi All,

    As suggested impala v1.0 has solved that issue I was facing. Thanks to
    all Cloudera's engineers for making it a hit.

    Regards,
    *Rahul Mathur*

    On Tue, Apr 30, 2013 at 11:23 AM, rahul mathur wrote:

    @Nong Li : thanx for sharing the issue, that is really helpful to
    know what was causing that error.
    @Lenni : Its hard for me to look into the C code as it is not my
    forte, Java is. Only option left with me is to wait for new release :).

    Thanks,
    *Rahul Mathur*

    On Tue, Apr 30, 2013 at 5:13 AM, Lenni Kuff wrote:

    This fix has been pushed to the Impala githup repo if you are
    interested in digging into the code:

    https://github.com/cloudera/**im**pala/commit/**07dc2a206bc762fe31**
    591a7748b180**604787baca<https://github.com/cloudera/impala/commit/07dc2a206bc762fe31591a7748b180604787baca>

    Thanks,
    Lenni

    On Mon, Apr 29, 2013 at 9:52 AM, Nong Li wrote:

    The issue is that we were using a hardware instruction (popcnt)
    without first checking if
    the chip supported it. The fix is simply to fall back to a less
    optimized implementation
    on older (pre-nehalem for intel, pre-barcelona for amd) machines.

    The 1.0 release is coming out soon and will contain this fix.


    On Mon, Apr 29, 2013 at 3:01 AM, rahul mathur <rahulmat...@gmail.com
    wrote:
    Hi lskuff,
    *
    *
    I was not in touch with my mails from past few days, Whereas I can
    see same logs as posted by* *宾莉金 at my end. I followed the mail
    trail and so is the solution suggested impala *V1.0*.
    I am waiting for its release impatiently now and hope to download
    it soon. Also, if any one could explain what the problem was and
    fix is, that will really be appreciated.

    Thanks in advance,
    Rahul Mathur
    *
    *
    On Friday, April 26, 2013 2:28:09 AM UTC+5:30, Aaron T. Myers wrote:

    Hi Rahul,

    Moving this over to impal...@cloudera.org, which should be able
    to help you with this issue.


    --
    Aaron T. Myers
    Software Engineer, Cloudera


    On Thu, Apr 25, 2013 at 6:05 AM, rahul mathur <
    rahulmat...@gmail.com> wrote:
    Hi,

    I am facing an error : "*Error communicating with impalad:
    TSocket read 0 bytes" *which causes a JRE crash, this error
    results when I run select query on impala-shell. Following are the logs
    generated after this error in impala_server.log :

    13/04/25 11:09:23 INFO service.Frontend: analyze query select *
    from ex
    13/04/25 11:09:23 INFO catalog.HdfsTable: load table ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: load partition block md
    for ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded partition
    PartitionBlockMetadata{#**blocks****=0, #filenames=0,
    totalStringLen=0}
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded partition
    PartitionBlockMetadata{#**blocks****=1, #filenames=1,
    totalStringLen=104}
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded disk ids for
    table default.ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: 2
    13/04/25 11:09:24 INFO catalog.HdfsTable: table #rows=0
    13/04/25 11:09:24 INFO service.Frontend: create plan
    13/04/25 11:09:24 INFO planner.Planner: create single-node plan
    13/04/25 11:09:24 INFO planner.HdfsScanNode: collecting
    partitions for table ex
    13/04/25 11:09:24 INFO planner.HdfsScanNode: finalize HdfsScan:
    cardinality=0
    13/04/25 11:09:24 INFO planner.HdfsScanNode: finalize HdfsScan:
    #nodes=2
    13/04/25 11:09:24 INFO planner.Planner: create plan fragments
    13/04/25 11:09:24 INFO planner.Planner: finalize plan fragments
    13/04/25 11:09:25 INFO service.Frontend: get scan range locations
    13/04/25 11:09:25 INFO service.Frontend: create result set
    metadata
    13/04/25 11:09:25 INFO service.JniFrontend: Plan Fragment 0
    UNPARTITIONED
    EXCHANGE (1)
    TUPLE IDS: 0

    Plan Fragment 1
    RANDOM
    STREAM DATA SINK
    EXCHANGE ID: 1
    UNPARTITIONED

    SCAN HDFS table=default.ex #partitions=1 size=265B (0)
    TUPLE IDS: 0

    #
    # A fatal error has been detected by the Java Runtime Environment:
    #
    # SIGILL (0x4) at pc=0x00000000008ed857, pid=22887,
    tid=140607123736320
    #
    # JRE version: 7.0_21-b11
    # Java VM: Java HotSpot(TM) 64-Bit Server VM (23.21-b01 mixed
    mode linux-amd64 compressed oops)
    # Problematic frame:
    # C [impalad+0x4ed857] * impala::HdfsScanNode::UpdateC
    ounters()+0x97*
    #
    # Failed to write core dump. Core dumps have been disabled. To
    enable core dumping, try "ulimit -c unlimited" before starting Java again
    #
    # An error report file with more information is saved as:
    # */var/run/impala/hs_err_pid22887.log*
    #
    # If you would like to submit a bug report, please visit:
    # http://bugreport.sun.com/**bugre****port/crash.jsp<http://bugreport.sun.com/bugreport/crash.jsp>


    There is one more log file generated under the home directory of
    impala user like "hs_err_pid22887.log" by JRE.
    I tried running same queries on HIVE which showed correct result
    from local and remote client. Also, tried both 0.7 and latest version 1.0
    but both are throwing same error. Please reply with a solution ASAP.

    Thanks in advance,
    *Rahul Mathur*

    --


  • Amigo at Jun 3, 2013 at 3:11 pm
    Here is the output of :25000; however I did realize that the port is not
    open externally. Although, I don't think the "debug web server" would have
    to be open for all, right?

    impalad version 1.0 RELEASE (build d1bf0d1dac339af3692ffa17a5e3fdae0aed751f)
    Built on Sun, 28 Apr 2013 15:59:46 PST

    Hardware Info

    Cpu Info:
       Model: Intel(R) Xeon(R) CPU E5-2660 0 @ 2.20GHz
       Cores: 32
       L1 Cache: 0.00
       L2 Cache: 0.00
       L3 Cache: 0.00
       Hardware Supports:
         ssse3
         sse4_1
         sse4_2
         popcnt
    Mem Info: 252.01 GB
    Disk Info:
       Num disks 12: cciss/c0d, cciss/c0d0p, sda, sdb, sdc, sdd, dm-, VxDMP, VxDMP2p, VxDMP3p, VxDMP4p, VxVM



    On Saturday, June 1, 2013 10:03:06 AM UTC-5, lskuff wrote:

    Thanks for the information. I don't see anything immediately obvious from
    the logs of the table definition. In your browser could you go to:
    http://<hostname>:25000 on one of the machines running Impalad? You will
    see a "Hardware Info" box, could you send that output back?

    Thanks,
    Lenni

    On Thu, May 30, 2013 at 9:08 AM, Amigo <nitin....@gmail.com <javascript:>>wrote:
    Hive output

    Logging initialized using configuration in
    jar:file:/opt/cloudera/parcels/CDH-4.2.1-1.cdh4.2.1.p0.5/lib/hive/lib/hive-common-0.10.0-cdh4.2.1.jar!/hive-log4j.properties
    Hive history file=/tmp/ME/hive_job_log_ME_201305301035_1001543262.txt

    OK

    first_visit_date string

    acquisition_source_url string

    first_url string

    ip_address string

    browser_os_name string



    Detailed Table Information Table(tableName:cs, dbName:mobile, owner:ME,
    createTime:1369863897, lastAccessTime:0, retention:0,
    sd:StorageDescriptor(cols:[FieldSchema(name:first_visit_date, type:string,
    comment:null), FieldSchema(name:acquisition_source_url, type:string,
    comment:null), FieldSchema(name:first_url, type:string, comment:null),
    FieldSchema(name:ip_address, type:string, comment:null),
    FieldSchema(name:browser_os_name, type:string, comment:null)],
    location:hdfs://foobar-1.com:8020/bi/mobile/cs,
    inputFormat:org.apache.hadoop.mapred.TextInputFormat,
    outputFormat:org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat,
    compressed:false, numBuckets:0, serdeInfo:SerDeInfo(name:null,
    serializationLib:org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe,
    parameters:{serialization.format= , field.delim=
    Table creation:
    drop table if exists cs;

    create external table cs

    (

    FIRST_VISIT_DATE string,ACQUISITION_SOURCE_URL string,DESTINATION_URL
    string,IP_ADDRESS string,BROWSER_OS_NAME string
    )

    row format delimited fields terminated by '\t'

    location '/mobile/cs';
    Logs I found in /var/logs/impalad are here - http://pastebin.com/BgtnRq7Y

    thanks, nitin

    On Thursday, May 30, 2013 10:19:41 AM UTC-5, lskuff wrote:

    Hi Nitin,
    Could you please provide the impalad logs so we can help investigate
    this problem? It would also be useful to have "describe extended" output of
    your target table. You can get this by running:
    hive -e "describe extended cs"
    Thanks,
    Lenni
    Software Engineer - Cloudera

    On Wed, May 29, 2013 at 9:11 PM, Amigo wrote:

    So, started a brand new installed using CDH 4.5.2

    *Version*: 4.5.2 (#327 built by jenkins on 20130429-1453 git:
    16cab2c7b76194b7877d64a4215494**daa387a266)
    impala-shell -v
    Impala Shell v1.0 (d1bf0d1) built on Sun Apr 28 15:33:52 PDT 2013

    Using IMPALA 1.0-1.p0.371 parcel

    - MR jobs work
    - HiveQL works (for same queries stated below)
    - Impala
    - Simple select works (in hive too)
    - Queries with group fails

    Query: select search_word, count(*) from cs group by search_word

    Error communicating with impalad: TSocket read 0 bytes



    Anybody spot what I may be doing wrong? Missing any config? I couldn't
    find anything useful in /var/logs/impala/impalad.INFO
    What else can I provide?

    thanks, nitin

    On Thursday, May 2, 2013 5:42:06 AM UTC-5, rahul mathur wrote:

    Hi All,

    As suggested impala v1.0 has solved that issue I was facing. Thanks to
    all Cloudera's engineers for making it a hit.

    Regards,
    *Rahul Mathur*

    On Tue, Apr 30, 2013 at 11:23 AM, rahul mathur wrote:

    @Nong Li : thanx for sharing the issue, that is really helpful to
    know what was causing that error.
    @Lenni : Its hard for me to look into the C code as it is not my
    forte, Java is. Only option left with me is to wait for new release :).

    Thanks,
    *Rahul Mathur*

    On Tue, Apr 30, 2013 at 5:13 AM, Lenni Kuff wrote:

    This fix has been pushed to the Impala githup repo if you are
    interested in digging into the code:

    https://github.com/cloudera/**im**pala/commit/**07dc2a206bc762fe31**
    591a7748b180**604787baca<https://github.com/cloudera/impala/commit/07dc2a206bc762fe31591a7748b180604787baca>

    Thanks,
    Lenni

    On Mon, Apr 29, 2013 at 9:52 AM, Nong Li wrote:

    The issue is that we were using a hardware instruction (popcnt)
    without first checking if
    the chip supported it. The fix is simply to fall back to a less
    optimized implementation
    on older (pre-nehalem for intel, pre-barcelona for amd) machines.

    The 1.0 release is coming out soon and will contain this fix.


    On Mon, Apr 29, 2013 at 3:01 AM, rahul mathur <
    rahulmat...@gmail.com> wrote:
    Hi lskuff,
    *
    *
    I was not in touch with my mails from past few days, Whereas I can
    see same logs as posted by* *宾莉金 at my end. I followed the mail
    trail and so is the solution suggested impala *V1.0*.
    I am waiting for its release impatiently now and hope to download
    it soon. Also, if any one could explain what the problem was and
    fix is, that will really be appreciated.

    Thanks in advance,
    Rahul Mathur
    *
    *
    On Friday, April 26, 2013 2:28:09 AM UTC+5:30, Aaron T. Myers
    wrote:
    Hi Rahul,

    Moving this over to impal...@cloudera.org, which should be able
    to help you with this issue.


    --
    Aaron T. Myers
    Software Engineer, Cloudera


    On Thu, Apr 25, 2013 at 6:05 AM, rahul mathur <
    rahulmat...@gmail.com> wrote:
    Hi,

    I am facing an error : "*Error communicating with impalad:
    TSocket read 0 bytes" *which causes a JRE crash, this error
    results when I run select query on impala-shell. Following are the logs
    generated after this error in impala_server.log :

    13/04/25 11:09:23 INFO service.Frontend: analyze query select *
    from ex
    13/04/25 11:09:23 INFO catalog.HdfsTable: load table ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: load partition block
    md for ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded partition
    PartitionBlockMetadata{#**blocks****=0, #filenames=0,
    totalStringLen=0}
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded partition
    PartitionBlockMetadata{#**blocks****=1, #filenames=1,
    totalStringLen=104}
    13/04/25 11:09:24 INFO catalog.HdfsTable: loaded disk ids for
    table default.ex
    13/04/25 11:09:24 INFO catalog.HdfsTable: 2
    13/04/25 11:09:24 INFO catalog.HdfsTable: table #rows=0
    13/04/25 11:09:24 INFO service.Frontend: create plan
    13/04/25 11:09:24 INFO planner.Planner: create single-node plan
    13/04/25 11:09:24 INFO planner.HdfsScanNode: collecting
    partitions for table ex
    13/04/25 11:09:24 INFO planner.HdfsScanNode: finalize HdfsScan:
    cardinality=0
    13/04/25 11:09:24 INFO planner.HdfsScanNode: finalize HdfsScan:
    #nodes=2
    13/04/25 11:09:24 INFO planner.Planner: create plan fragments
    13/04/25 11:09:24 INFO planner.Planner: finalize plan fragments
    13/04/25 11:09:25 INFO service.Frontend: get scan range locations
    13/04/25 11:09:25 INFO service.Frontend: create result set
    metadata
    13/04/25 11:09:25 INFO service.JniFrontend: Plan Fragment 0
    UNPARTITIONED
    EXCHANGE (1)
    TUPLE IDS: 0

    Plan Fragment 1
    RANDOM
    STREAM DATA SINK
    EXCHANGE ID: 1
    UNPARTITIONED

    SCAN HDFS table=default.ex #partitions=1 size=265B (0)
    TUPLE IDS: 0

    #
    # A fatal error has been detected by the Java Runtime
    Environment:
    #
    # SIGILL (0x4) at pc=0x00000000008ed857, pid=22887,
    tid=140607123736320
    #
    # JRE version: 7.0_21-b11
    # Java VM: Java HotSpot(TM) 64-Bit Server VM (23.21-b01 mixed
    mode linux-amd64 compressed oops)
    # Problematic frame:
    # C [impalad+0x4ed857] * impala::HdfsScanNode::UpdateC
    ounters()+0x97*
    #
    # Failed to write core dump. Core dumps have been disabled. To
    enable core dumping, try "ulimit -c unlimited" before starting Java again
    #
    # An error report file with more information is saved as:
    # */var/run/impala/hs_err_pid22887.log*
    #
    # If you would like to submit a bug report, please visit:
    # http://bugreport.sun.com/**bugre****port/crash.jsp<http://bugreport.sun.com/bugreport/crash.jsp>


    There is one more log file generated under the home directory of
    impala user like "hs_err_pid22887.log" by JRE.
    I tried running same queries on HIVE which showed correct result
    from local and remote client. Also, tried both 0.7 and latest version 1.0
    but both are throwing same error. Please reply with a solution ASAP.

    Thanks in advance,
    *Rahul Mathur*

    --


  • Screenthong Lapmahapaisan at Aug 2, 2013 at 8:53 am
    Hi lskuff,

         I face the same issue on cloudera impala 1.1. I 'm runnning "set
    disable_codegen=true" impalad did not died.

    I have generaged core from impalad , If you need to find out root cause.


    [core file ]
    http://www.boyr.com/getfile.php?id=1402049&key=51fb6cc71b178 (plase
    rename to impala.tar.gz)

    or

    http://www.speedyshare.com/a5UdG/impala.tar.gz

    hive> describe extended proxy_snappy_seq;
    OK
    date_time string
    category string
    msisdn string
    client_ip string
    http_code string
    content_length int
    method string
    domain string
    dest_port string
    uri string
    content_type string
    uaprofile string
    dest_ip string
    server_ip string

    Detailed Table Information Table(tableName:proxy_snappy_seq,
    dbName:ais, owner:root, createTime:1375430008, lastAccessTime:0,
    retention:0, sd:StorageDescriptor(cols:[FieldSchema(name:date_time,
    type:string, comment:null), FieldSchema(name:category, type:string,
    comment:null), FieldSchema(name:msisdn, type:string, comment:null),
    FieldSchema(name:client_ip, type:string, comment:null),
    FieldSchema(name:http_code, type:string, comment:null),
    FieldSchema(name:content_length, type:int, comment:null),
    FieldSchema(name:method, type:string, comment:null),
    FieldSchema(name:domain, type:string, comment:null),
    FieldSchema(name:dest_port, type:string, comment:null),
    FieldSchema(name:uri, type:string, comment:null),
    FieldSchema(name:content_type, type:string, comment:null),
    FieldSchema(name:uaprofile, type:string, comment:null),
    FieldSchema(name:dest_ip, type:string, comment:null),
    FieldSchema(name:server_ip, type:string, comment:null)],
    location:hdfs://hadoop.g-able.com:8020/user/hive/warehouse/ais.db/proxy_snappy_seq,
    inputFormat:org.apache.hadoop.mapred.SequenceFileInputFormat,
    outputFormat:org.apache.hadoop.hive.ql.io.HiveSequenceFileOutputFormat,
    compressed:false, numBuckets:-1, serdeInfo:SerDeInfo(name:null,
    serializationLib:org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe,
    parameters:{serialization.format=|, line.delim=
    , field.delim=|}), bucketCols:[], sortCols:[], parameters:{},
    skewedInfo:SkewedInfo(skewedColNames:[], skewedColValues:[],
    skewedColValueLocationMaps:{}), storedAsSubDirectories:false),
    partitionKeys:[], parameters:{numPartitions=0, numFiles=18,
    transient_lastDdlTime=1375430448, totalSize=5101757765, numRows=0,
    rawDataSize=0}, viewOriginalText:null, viewExpandedText:null,
    tableType:MANAGED_TABLE)

    impala-shell> select msisdn from proxy_snappy_seq group by msisdn limit 10;
    Query: select msisdn from proxy_snappy_seq group by msisdn limit 10
    Query finished, fetching results ...
    Error communicating with impalad: TSocket read 0 bytes
    Cancelling query ...
    Not connected (use CONNECT to establish a connection)
    [Not connected] >


    Regards,
    Screen
  • Screenthong Lapmahapaisan at Aug 2, 2013 at 11:33 am
    More information,

       I use RHEL 5.8. I found some case like this use RHEL 5.9. May be RH
    version has to concern.
  • Lenni Kuff at Aug 2, 2013 at 3:00 pm
    Hi,
    We have identified and fixed the problem (see
    https://issues.cloudera.org/browse/IMPALA-477) and the fix will be
    available in the next Impala release. It is a problem related to codegen
    and RHEL5, so until the next release goes out the workarounds are:

    1) Disable codegen
    2) Upgrade to RHEL6

    We understand that neither of these options are ideal and will prioritize
    getting a fix available as soon as possible.

    Thanks,
    Lenni
    On Fri, Aug 2, 2013 at 4:33 AM, Screenthong Lapmahapaisan wrote:


    More information,

    I use RHEL 5.8. I found some case like this use RHEL 5.9. May be RH
    version has to concern.

Related Discussions

Discussion Navigation
viewthread | post
Discussion Overview
groupimpala-user @
categorieshadoop
postedApr 25, '13 at 8:58p
activeAug 2, '13 at 3:00p
posts14
users6
websitecloudera.com
irc#hadoop

People

Translate

site design / logo © 2022 Grokbase