manage-02 fails to index certain headings; ue_01 *does* index them
- Product: Aleph
- Product Version: 20, 21, 22, 23
- Relevant for Installation Type: Dedicated-Direct, Direct, Local, Total Care
Description:
Though manage-02 running for the xxx10 authority library runs to completion and indexes most of the headings, it fails to index certain headings.
When the authority record containing such an unindexed heading is updated in GUI Cataloging and reindexed by ue_01, the missing headings *are* indexed correctly.
Resolution:
The $alephe_scratch xxx10_p_manage_02 job log contains the following error:
in_fl: /exlibris/aleph/u20_2/xxx10/files/z01.seq out_fl: /exlibris/aleph/u20_2/xxx10/files/z01.comp.seq
Segmentation fault
and the ./xxx10/files/z01.seq file is larger than 2 Gig.
There can be a problem with work files larger than 2 Gig. The solution is to decrease the value of the p_manage_02_loop_length parameter in the ./xxx10/prof_library from 50000 to 40000.
Additional Info
The p_manage_02_loop_length controls the number of records included in each work file the job constructs. Generally, the larger the loop-length, the faster the job will run. But the work file can't include so many records that it will exceed two gigabtyes in size (-- which is what happened in this case). Generally, we have found that 50,000 is a good number. But in several cases the p_manage_01_loop_length needed to be reduced from 50000 to 40000. And now in this case the p_manage_02_loop_length.
The headings which were missing from the index were ones which were included in this problem cycle (the z01.seq). Headings which were in other cycles had no problem and were processed successfully. It seems that the records in this range had more headings to be indexed, resulting in the larger work file.
If this problem is encountered with the manage-02, it could happen with manage-01 as well. Thus, we suggest changing the xxx10 prof_library p_manage_01_loop_length to 40000 also in this case.
- Article last edited: 02-Mar-2016