Skip to main content
ExLibris
  • Subscribe by RSS
  • Ex Libris Knowledge Center

    ALEPH 18 - Oracle handle_connection error; TWO_TASK problems

    • Article Type: General
    • Product: Aleph
    • Product Version: 18.01

    Description:
    We are trying to get going on our v.17 to v.18 upgrade and we have hit a problem. This happened at the point we were going to run the imp_demo_library command. We are getting this kind of error running the command and when we login as the v.18 ALEPH user and dlib:

    oit.abc.edu-a18(1) >>dlib abc01
    --------------------------------------------------------------
    Load: /exlibris/aleph/u18_1/alephe/tab/tab100
    --------------------------------------------------------------
    Load: /exlibris/aleph/u18_1/abc01/tab/tab100
    ld.so.1: oracle: fatal: libskgxp10.so: open failed: No such file or directory

    Oracle error: handle_connection
    ORA-12547: TNS:lost contact

    Our DBA has looked at this and thinks the problem is that v.18 is not really "seeing" the Oracle 10 database. Our Oracle path doesn't match the default, our is at: /oracle/u01/app/oracle/product/10.2.0
    The DBA said he changed some symbolic links that were pointing to the default installation. We've also tried some changes to aleph_start. The one we used with our first try of imp_demo_library was the one now named:

    aleph_start_20060524jma_noLD_PATH_64

    At that point the imp_demo_library errors also included errors like this:

    Starting import of library USM01
    Tue May 30 14:39:12 CDT 2006
    create_ora_user_b usm01
    ld.so.1: sqlplus: fatal: /oracle/u01/app/oracle/product/10.2.0/db/lib32/libsqlplus.so: wrong ELF class: ELFCLASS32
    Killed
    cat: cannot open /exlibris/aleph/a18_1/usm01/scratch/tablespaces.lst
    ld.so.1: sqlplus: fatal: /oracle/u01/app/oracle/product/10.2.0/db/lib32/libsqlplus.so: wrong ELF class: ELFCLASS32
    Killed

    USM01 table all
    l_op == table
    l_table == all
    ld.so.1: sqlplus: fatal: /oracle/u01/app/oracle/product/10.2.0/db/lib32/libsqlplus.so: wrong ELF class: ELFCLASS32
    Killed
    ==============
    Then we added a LD_LIBRARY_PATH_64 line to aleph_start, like this:
    setenv LD_LIBRARY_PATH_64 "$ORACLE_HOME/lib:/usr/ccs/bin:/usr/lib:/usr/ucb/lib:/
    usr/ucblib:$ORACLE_HOME/ctx/lib"
    for the case SUN_OS_2:

    This or other things our DBA did, took care of the "ELFclass" errors.

    Our DBA has made further changes to aleph_start, but none of then have eliminated the handle_connection message. We really want to get going on our first run of Upgrade Express.

    --------------------------------------------------------------------------------------------------


    In response to your questions ...

    1) The ORACLE_SID looks okay:

    oit.abc.edu-18(1) ABC01-ALEPH>>echo $ORACLE_SID
    aleph18

    2) Never looked at this env. variable, should it look like this?

    oit.abc.edu-18(1) ABC01-ALEPH>>echo $aleph_db
    saturn.aleph18

    3) the s+ aleph_admin gets me an error:

    oit.abc.edu-18(1) ABC01-ALEPH>> s+ aleph_admin

    SQL*Plus: Release 10.2.0.1.0 - Production on Fri Jun 2 14:07:59 2006

    Copyright (c) 1982, 2005, Oracle. All rights reserved.

    ld.so.1: oracle: fatal: libskgxp10.so: open failed: No such file or directory
    ERROR:
    ORA-12547: TNS:lost contact

    Enter user-name:

    4) I will check with our DBAs with respect to the oracle users and let you know.

    Resolution:
    I copied version 18 aleph_start to aleph_start_20060605_eyal and added to aleph_start the environment variable
    TWO_TASK: –

    setenv TWO_TASK ${ORA_HOST}.${ORACLE_SID}

    Now it is ok.

    [Eyal Alkalay]


    • Article last edited: 10/8/2013