Archive logs filling up; ue_01 cycling between two records repeatedly
- Article Type: General
- Product: Aleph
- Product Version: 20
Description:
The p_manage_18 job "completed" but ever since that time, something has been running on the server, related to ABC10 indexing, which is just running continuously. I don't know what it is and need help in figuring it out.
It has to do with ABC10 indexing because:
a) If I turn off indexing, the Oracle logging stops so processing halts.
b) z105 records stop getting created in the $z105_library, whereas when it is running, such records are created.
Something may be legitimately running, but it *seems* that I have runaway processing going on . It looks more and more, after two full days of running, that something is not working as expected.
Resolution:
Looking at the abc10 $data_scratch/run_e_01.nnnn log, I see that it is cycling between two records over and over:
HANDLING DOC NO. - ABC10.000303937 09:09:09
HANDLING DOC NO. - ABC10.001058015 09:09:10
HANDLING DOC NO. - ABC10.000303937 09:09:10
HANDLING DOC NO. - ABC10.001058015 09:09:10
HANDLING DOC NO. - ABC10.000303937 09:09:11
HANDLING DOC NO. - ABC10.001058015 09:09:11
HANDLING DOC NO. - ABC10.000303937 09:09:12
HANDLING DOC NO. - ABC10.001058015 09:09:12
HANDLING DOC NO. - ABC10.000303937 09:09:12
HANDLING DOC NO. - ABC10.001058015 09:09:13
I am unable to look at these records in util f/4 because they have too many CAT fields.
The temporary solution was to:
* stop the abc10 ue_01 process
* stop the $z105_library ue_11 process
* stop the bib library ue_08 process
* delete the abc10 z07 records for each of these abc10 doc records
* use this SQL to locate their z105 records:
abc50@ALEPH20> select count(*) from z105 where z105_text like 'ABC10:000303937%';
**** Hit return to continue ****
COUNT(*)
----------
1456
1 row selected.
abc50@ALEPH20> select count(*) from z105 where z105_text like 'ABC10:001058015%';
**** Hit return to continue ****
COUNT(*)
----------
1458
and delete those as well.
Then restart the ue processes which were stopped. (See above.)
Something will need to be done about abc10 doc records 001058015 and 000303937. It may be just one is the problem, but it seems to me better to consider both problematic.
They currently have too many CAT fields to be viewable. Maybe you can bring them up in the GUI and delete the CAT fields(?)
If not, I suggest that you run the p_manage_21 global change for ABC10 with an input file consisting of:
001058015ABC10
000303937ABC10
p_manage_21 has an option to delete fields. You should delete the CAT fields.
Then let me know and I will look at these records.
- Article last edited: 10/8/2013