974670 wrote:That would have been my first idea for the scenario you describe. How did you change the table and indexes to use a higher initrans value?
... We increased INITRANS and rebuilt the tables and indexes on them (with higher value than DOP used in DML).
operation (or drop and recreate).
alter table move tablespace
1) Is the statement type listed above a valid statement which should run correctly?Do you see already the wait events and what kind of waits do you experience? An ITL shortage would be something like an TX 4 wait if I remember correctly.
2) If it is a valid statement are there any other situations which can cause deadlocks (except ITL shortage, missing indexes on FKs, conflicting transactions)?
3) Are there any oracle bugs related to deadlocks which may apply here? (I did some search online, however haven't found anything interesting)
4) Can we turn on more tracing to capture information which could help solve the riddle? If yes what would that tracing be?