On Wed, Nov 6, 2019 at 2:15 AM Michael Paquier wrote:
> On Tue, Nov 05, 2019 at 01:19:37PM +0900, Michael Paquier wrote:
> > On Sat, Nov 02, 2019 at 09:54:54PM -0400, Jeff Janes wrote:
> >> Filling out the timestamp after the message has already been sent is
> taking
> >> "as late as possible" a
On Tue, Nov 05, 2019 at 01:19:37PM +0900, Michael Paquier wrote:
> On Sat, Nov 02, 2019 at 09:54:54PM -0400, Jeff Janes wrote:
>> Filling out the timestamp after the message has already been sent is taking
>> "as late as possible" a little too far. It results in every message having
>> a zero time
On Sat, Nov 02, 2019 at 09:54:54PM -0400, Jeff Janes wrote:
> While monitoring pg_stat_subscription, I noticed that last_msg_send_time
> was usually NULL, which doesn't make sense. Why would we have a message,
> but not know when it was sent?
So... The timestamp is received and stored in Logical
While monitoring pg_stat_subscription, I noticed that last_msg_send_time
was usually NULL, which doesn't make sense. Why would we have a message,
but not know when it was sent?
Looking in src/backend/replication/walsender.c, there is this:
/* output previously gathered data in a CopyData pac
