On Wed, 2015-12-02 at 10:47 -0800, Christophe Pettus wrote:
> On Dec 2, 2015, at 10:29 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> > The short answer is that heavyweight tuple locks can be taken
> > internally
> > by UPDATE, DELETE, and other row-level operations, and the
> > specifics of
> > which type of lock any given action takes are implementation
> > details that
> > can change from time to time.
>
> Great, thank you!
>
> --
> -- Christophe Pettus
> xof@thebuild.com
>
>
>
Don't know if this helps at all. On the hot standby doco.
http://www.postgresql.org/docs/9.4/static/hot-standby.html
About two-thirds of the way down it refers to the pg_locks table.
Are you using "hot standby"?