On Wed, Dec 14, 2016 at 8:38 AM, Robert Haas <robertmh...@gmail.com> wrote:
> On Wed, Dec 14, 2016 at 9:26 AM, Kevin Grittner <kgri...@gmail.com> wrote:

>> Predicate locks
>> from reads within subtransactions are not discarded, even if the
>> work of the subtransaction is otherwise discarded.
> Oh, interesting.  Just to be clear, I'm not lobbying to change that; I
> was guessing (very late at night) what decision you probably made and
> it seems I was incorrect.  But doesn't that imply that if a read fails
> in a subtransaction with serialization_error, the parent MUST also be
> killed with serialization_error?

To prevent serialization anomalies, a top level transaction which
gets a serialization error in a subtransaction must fail.  To
provide the best information for an application framework which
wants to make smart decisions about when to retry a transaction
from the start, it should fail with a serialization error.  I'm
starting to think that, in addition to the doomed flag mechanism,
we should perhaps (as you suggested earlier in this thread) not
allow the serialization failure exception to be caught.  Or require
that it be re-thrown?  I don't know.  Maybe if someone catches a
serialization failure error, they should just be told that they
can't complain about where, later in the transaction, it might be
re-thrown.  (And yes, that could be in a COMMIT, even for a
read-only transaction.)  It may be that catching a serialization
failure and throwing a *different* error could confuse the
application framework's retry logic; I think at that point we have
to give up and let that be.  There may be some circumstances where
there is a valid need to replace a serialization failure with some
other error that you *want* to have appear in front of an end user.

>> Fortunately, Thomas Munro took an interest in the problem as it
>> related to duplicates on primary keys, unique constraints, and
>> unique indexes, and put forward a patch that cured the defect in
>> the common cases, and provided an easy workaround for the one case
>> he was unable to fix in that initial patch.  To finish the work for
>> these constraints and indexes, I think we need to add predicate
>> locking while descending to the insertion point  during the check
>> for an existing duplicate.
> I suggest adding something about this to README-SSI as a known issue.

Good idea.  Will do.

Kevin Grittner
EDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:

Reply via email to