Re: Sending unflushed WAL in physical replication

Hi Rahila,

> Please find attached a POC patch that introduces changes to the WAL sender and
> receiver, allowing WAL records to be sent to standbys before they are flushed
> to disk on the primary during physical replication. [..]

I didn't look at the code but your description of the design sounds OK.

I wanted to clarify: what happens if master doesn't increase flushPtr
and replica runs out of memory for WAL records?

> Benchmark details are as follows:
> Synchronous replication with remote write enabled.
> Two Azure VMs: Central India (primary), Central US (standby).
> [...]

I'm curious what happens:

1. When master and replica are located in the same datacenter.
2. What happens for small transactions?

--
Best regards,
Aleksander Alekseev