On Wed, Feb 9, 2011 at 7:12 AM, Itagaki Takahiro
<itagaki.takahiro@gmail.com> wrote:
> One issue might be in pg_locks, as you pointed out in the previous mail:
>> if a session holds both a transaction level and a session level lock
>> on the same resource, only one of them will appear in pg_locks.
> Also, we cannot distinguish transaction-level locks from session-level
> locks from pg_locks.
>
> It was not an issue before because session locks are only used in
> internal implementation. It looks as a transaction from users.
> However, this feature reveals the status in public. We might need
> to add some bits to shared lock state to show which lock is session-level.
Presumably that would carry a small performance penalty, since
changing the status of the lock would require modifications to the
shared hash table, not just the backend-private one.
It may still be worth doing, but I'm inclined to think that it's a
separate patch that someone could submit for 9.2.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company