manage-01 in parallel library extremely slow

• Article Type: General
• Product: Aleph
• Product Version: 20
• Relevant for Installation Type: Dedicated-Direct; Direct; Local;

p_manage_01 has been running in the abc09 parallel bib library.

Resolution:
If you don't already have it, you should get the "How To Run Index Jobs" document (on the Doc Portal: Tree Search > How To from Support Department > Indexing_filing_and_expand_procedures / How_To_Run_Index_Jobs_2009_Nov_02_js.doc .)

Section 10 has run times. 10 hours for each million bib records typical. Since the job has been running on abc09 for six days, clearly something is wrong.

As described in the "Job is Slow" topic in Section 9 (Troubleshooting), one of the main causes of slowness is missing Oracle indexes on the tables being read. The main table being read by this job is the z00. util a/17/14) shows that the following Oracle indexes are missing: z00_id, z103_id, z103_id1.

We see (in abc09 util a/17/5/1) that there are synonyms pointing to abc01 for the z00 and z103 tables. But the abc09 file_list has no "LS" entries. util a/17/5/2 ("Create/Recreate All Library Synonyms") reads the file_list, looking for "LS" entries. So, it seems these synonyms were created by some method *other than* util a/17/5/2 -- and that that method failed to create synonyms for the z00_id, z103_id, and z103_id1.

Please:

1. Kill the abc09 p_manage_01

2. Get the "How To Run Index Jobs" and "Parallel Indexing" documents.

3. As described in the Parallel Indexing, add LS's for z00, z103, z00_id, z103_id, and z103_id1 to the abc09 file_list.

4. Do util a/17/5/2 to create the proper synonyms.

5. Rerun the abc09 p_manage_01.