Simon Riggs <simon@2ndQuadrant.com> wrote:
> I foresee a function that tells you the delay based on a protocol
> message of 'k' for keepalive.
If the delay you mention is basically a "ping" time or something
similar, that would answer the need I've been on about. We need to
know, based on access to the replica, that the replication system is
alive and well even with an idle master; as it can otherwise be hard
to distinguish an idle master from a failed replication system
(broken connection, misconfigured replication, etc.).
Right now there is a periodic checkpoint which flows through the WAL
and affects the pg_controldata report on the replica -- we've been
using that for monitoring. Any sort of heartbeat or ping which
provides sign-of-life on the connection, accessible on the replica,
should do for our purposes. If it only works over streaming
replication on a hot standby, that's OK -- we plan to be running
everything that way before 9.2 comes out, as long as we can
materialize "traditional" WAL files on the receiving end from the SR
stream.
-1 on changing the checkpoint behavior before 9.2, though.
-Kevin
