Hello!
In our production system with tens of thousands PostgreSQL clusters we
encounter exactly the same issue and forced to synchronize upstreams and
downstreams via external means, which is quite suboptimal.
I`ve done done some work on proposed patch would like to present it for a
discussion.
There are number of changes, such as sending just TLI and Segno instead of
full WAL filename, shifting some work into archiver and adding shared
memory for walreceiver/archiver synchronization.
There is number of issues currently unresolved, which I would also like to
discuss.
1. Should we update pg_stat_archiver on standby to support cascading
replication or should we just resend report, received from upstream?
Personally I incline more toward pg_stat_archiver path, because there will
be less `if-else` programming.
2. What should we do with *.history.ready, *.backup.ready and
.partial.ready files on standby? I think, we can just stamp them with .done.
3. Should we keep this awkward part with switchpont calculation in timeline
switch case? I think, all segments, that are not in our server history
should just be stamped with .done.
4. Currently .done is forced either by walreceiver (on receiving report
from upstream) and archiver. Should we move this into the archiver entirely?
Thank you for your interest in this topic!