On Wed, Feb 7, 2018 at 9:47 PM, Robert Haas <robertmh...@gmail.com> wrote:
> On Mon, Jan 22, 2018 at 7:05 AM, Amit Kapila <amit.kapil...@gmail.com> wrote:
>> On error, workers should be terminated. What kind of problem do you
>> have in mind?
> Hmm. Yeah, I guess that makes sense. If the only thing you can do is
> fetch from the cursor -- and you have to make sure to lock down
> protocol messages as well as SQL commands -- and if any error kills
> the workers -- then I guess this might be workable. However, I think
> there are still problems. Say the worker hits an error while the
> leader is idle; how do we report the error?
I guess workers need to wait till leader become active and processes
the error message.
> It's a protocol version
> for the leader to send an unsolicited ErrorResponse, which is why we
> have to use FATAL rather than ERROR in various places for recovery
> conflicts, query cancellation, etc.
> Also, if you're OK with not being able to do anything except fetch
> from the cursor until you reach the end, then couldn't you just issue
> the query without the cursor and use PQsetSingleRowMode?
I think there is a lot of cursor usage via plpgsql in which it could be useful.