Grigory Smolkin <g.smolkin@postgrespro.ru> wrote on 06/26/2019 12:37:10 PM:
> From: Grigory Smolkin <g.smolkin@postgrespro.ru>
> To: pgsql-general@lists.postgresql.org
> Date: 06/26/2019 12:37 PM
> Subject: [EXTERNAL] Re: Enabling checksums on a streaming replica
>
> Hello!
> Very interesting idea, but what about full page writes, that comes
> from master?
> Can they be a problem?
In testing, it doesn't appear to matter. I've ensured that I've generated some full page writes (confirmed via pg_waldump), and those apply fine.
The one thing I'm not sure of, when verifying checksums via pg_checksums I see blocks being skipped. I'm not sure what or why it is skipping blocks, and if that indicates a problem or not.
pg_checksums -c /data/10_repl/
Checksum scan completed
Files scanned: 1530
Blocks scanned: 100782
Blocks skipped: 3
Bad checksums: 0
Data checksum version: 1
Brad