Gregory Stark <stark@enterprisedb.com> writes:
> "Pavan Deolasee" <pavan.deolasee@gmail.com> writes:
>> Alternatively, we can just acquire AccessExclusiveLock on the main relation
>> before proceeding with the recursive deletion. That would solve this case,
>> but may be there are other similar deadlocks waiting to happen.
> Surely we should be locking the relation before even doing the dependency scan
Yeah. I think this is just another manifestation of the problem I was
noodling about a few days ago:
http://archives.postgresql.org/pgsql-hackers/2008-05/msg00301.php
As I said then, I don't want to think about it until after commitfest.
I foresee an invasive and not sanely back-patchable patch.
regards, tom lane