Gary_Cherlet

Corrupted OM index with bad owner pointers in member records

Discussion created by Gary_Cherlet on Jun 28, 2012
Latest reply on Jun 30, 2012 by Gary_Cherlet
Here is the history of this problem:

1) On a Sunday, 6/17, we did an unload/reload of the areas involved (along with some other areas not involved) to increase space.

2) Everything ran fine on Monday, 6/18.

3) On Tuesday, 6/19, we started seeing 1143 abends (corrupted index) and
DC061001 abends (which can also be caused by a corrupted index). The number of these abends slowly increased over the following week.

4) The abends were determined to caused by a corrupted OM index. In all cases examined, the member record had an owner pointer to the index, but the index did not have a member pointer to the record. The "disconnect from index" command failed with an 1126 (rec or index entry not found) for these records.

5) On Tuesday morning, 6/26, we ran a "maintain index" with "delete index"
to delete the entire index and zero out the bad pointers in the member records.

6) Everything ran fine all day Tuesay.

7) On Wednesday, 6/27, we started seeing the 1143 abends again.

8) On Thursday, 6/28 (today), we started seeing the DC061001 abends again.

9) The cause of the abends appears to be exactly the same as before.


The owners of the database say that they had no local mode abends, and did no restores, since the Tuesday morning fixup. So we are at a loss to explain how the index is getting corrupted. I have two questions:

a) Any ideas on how the index could be getting corrupted?

b) Is there any way to zero out the bad pointers in the member records, other than the "delete index" utility, which requires us to to delete the entire index? (Neither of the rebuild options will do this. We
tried.)


Kay Rozeboom

Outcomes