If some tool didn't find any problems doesn't mean that table/index is not corrupted. Did you try to delete and recreate indexes when it happen and see where the new record is located after that?
>That's what I initially thought too until I reproduced in multiple other data sets. I double-checked by running Abri data recover tools on the databases and no issues were found.
>
>This bug only happens intermittently. One record may commit fine, but the next one fails.
>
>No pending transactions... no anything. Just this sinking feeling that it's yet another bug related to tables being opened multiple times with the AGAIN ALIAS option.
--sb--