On Thu, Nov 01, 2018 at 04:44:40PM +0530, Amit Kapila wrote:
> This sounds like a good argument for having a whitelist approach, but
> is it really a big problem if a user gets warning for files that the
> utility is not able to verify checksums for? I think in some sense
> this message can be useful to the user as it can allow him to know
> which files are not verified by the utility for some form of
> corruption. I guess one can say that users might not be interested in
> this information in which case such a check could be optional as you
> seem to be suggesting in the following paragraph.
The replication protocol supports NOVERIFY_CHECKSUMS to avoid the
warnings so they enabled by default, and can be disabled at will.
pg_basebackup supports the same interface.
--
Michael