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

    p_manage_01_e "set: Syntax error"

    • Article Type: General
    • Product: Aleph
    • Product Version: 20

    Description:
    In checking the logs after the completion of keyword indexing on our test server, I see a number of "set: Syntax error" entries.

    I was able to check the ones in the p_manage_01_d1.log, using the information in KB 5719, i.e., there are no documents in the z00 for numbers 003840001 through 004000000 (for four cycles at 40,000 per cycle).

    But I can't figure out how to check the errors in the p_manage_01_e.log and p_manage_01_e2.log. There are z98 record in the one cycle I checked (cycle 72). Here is a snippet from the p_manage_01_e2.log (though not sure it includes all the necessary lines).

    Username:
    SQL*Loader: Release 11.1.0.7.0 - Production on Thu Dec 30 15:57:59 2010

    Copyright (c) 1982, 2007, Oracle. All rights reserved.


    Load completed - logical record count 0.
    Thu Dec 30 15:57:59 CST 2010
    set: Syntax error
    Cycle : 72
    Thu Dec 30 15:58:00 CST 2010
    Thu Dec 30 15:58:00 CST 2010
    >>>>>>>>>>>>> aleph_load_z9
    table_type = VAR load_mode = DIRECT option = go
    Loading Table : z98 table_type = VAR option=go
    *******table_type == VAR ==**************************

    =============== end of snippet =================


    When I looked up the z98 using the numbers in the cycles file, there were z98's in that range (and it wouldn't make sense for the job to leave skip ranges in the z98):
    SQL> select count (*) from z98 where z98_rec_key > '042023001' and z98_rec_key < '047472200';

    COUNT(*)
    ----------
    12256764

    I suspect I'm not looking in the right place.

    We were hitting 99% full in the abc01/scratch file system at one point and I moved 17G of manage_01_6 files to /wrkspc1/exlibris/aleph/u20_1/abc01/scratch (based on KB 5937).

    Resolution:
    Since the p_manage_01_e.cycles shows cycles 71, 72, and 73 with a "+" -- even though they get the "Syntax error" -- and since the job runs to completion, I tend to think that this is not really a problem....

    You write "There are z98 records in the one cycle I checked (cycle 72)....

    "When I looked up the z98 using the numbers in the cycles file, there were z98's in that range (and it wouldn't make sense for the job to leave skip ranges in the z98):
    SQL> select count (*) from z98 where z98_rec_key > '042023001' and z98_rec_key < '047472200';

    COUNT(*)
    ----------
    12256764 "


    It seems to me that zero z98 records were loaded in cycles 71, 72, and 73.

    Note: there is no way to check the input files to p_manage_01_e (/exlibris/aleph/u20_1/abc01/scratch/manage_01_6.nn) since the job deletes them. b_manage_01_4.cbl generates the manage_01_6.nn files by uncompressing other files.

    As can be seen in the ./alephm/source/Z98, the Z98-REC-NUMBER is bytes 4-12 in the z98_rec_key:

    02 Z98-REC-KEY.
    03 Z98-FIELD-NUMBER PICTURE 9(3).
    03 Z98-REC-NUMBER PICTURE 9(9).
    03 Z98-CONT PICTURE 9(9).


    And we see that there are no z98 records with Z98-REC-NUMBER's in this range:

    SQL> select count (*) from z98 where substr(z98_rec_key,4,9) > '042023001' and substr(z98_rec_key,4,9) < '047472200';


    COUNT(*)
    ----------
    0

    The site tried re-starting p_manage_01_e but had a different problem. (See KB 16384-34338.)

    They then made sure that they had sufficient room in the abc01 $data_scratch and re-ran p_manage_01 from the start, with the same parameters as they did originally. That worked.


    • Article last edited: 10/8/2013
    • Was this article helpful?