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

Description:
We wanted to deletd 126,748 abc60 HOL records and ran manage-18 with parameter "Delete a Record".

Looking at the logfile, it seems that p_manage_18 suppressed only 5,734 HOLs records. Any idea why manage-18 stop deleting HOLs records after 5,734?

After this, we are unable to update/create abc60 records in GUI Cataloging and the abc60 ue_01 stalls without processing any records.

Resolution:
The addition of diagnostics to the ./com/create_hol_record program indicated a problem with the abc60 z13/z13u table. I tried to do util a/17/1 on z13/z13u, in preparation for running p_manage_07 to rebuild the tables, but I got the following errors:

aleph_admin@ALEPH20> CREATE UNIQUE INDEX UQM60.Z13_ID ON ABC60.Z13 (Z13_REC_KEY ASC)
*
ERROR at line 1:
ORA-00054: resource busy and acquire with NOWAIT specified or timeout expired

This is like KB 8192-7896 and, as in that case, required stopping/restarting Oracle -- after which, everything was fine. Subsequent p_manage_18 runs got this same problem ... and, also, could be corrected only by restarting Oracle.

I suggest using p_manage_21 to add a "DEL Y" field to each HOL record rather than deleting them with p_manage_18. The use of this technique to delete ADM records is described in detail in KB 16384-32913, but, as noted in that record, the same procedure can be used for the xxx60 HOL records.

Note: Site found that stopping ue_01 while p_manage_18 or p_manage_21 was running prevented this problem.

[From site:]

I ran manage-21 without background indexing (ue_01) and it works fine : I did suppress 126 665 HOLs record without any problems at all. I also did a manage-21 (on 74 655 BIB records) in uqm01 that add a STA field and this time I didn't stop background indexing. No problem. In the future, we will stop background indexing when doing a big job in abc60.

See also KB 16384-38594.

keyword: deadlock

• Article last edited: 10/8/2013