the way reorg is currently is designed is that it operates as a single transaction. That means, all the work that reorg does is undone if the log is filled up (or any other error) and no space is available to extend the log.
Feature request #1 (design bug):
The granularity of all the reorg operations must be significantly increased. Perhaps at the X number of extents, consider that a ‘sub tran’ that will be committed even if the rest of the reorg fails due to space or other ‘non-critical’ issue.
- Sybase Feature Request (to be added)
- ISUG Feature request