I wrote:
> Please note also that what pgpool users have got right now is a time
> bomb, which is not better than immediately-visible breakage.
BTW, so far as that goes, I suggest that we tweak nextval() and setval()
to force the sequence tuple's xmax to zero. That will provide a simple
recovery path for anyone who's at risk at the moment. Of course, this
has to go hand-in-hand with the change to forbid SELECT FOR UPDATE,
else those operations would risk breaking active tuple locks.
regards, tom lane