David Steele

Sad to say, that would break backward compatibility. We don't look at the extension on restore because there could be files in PGDATA that end in .gz or .lz4 or...

I think if we were going to do anything here, it would be to add a command to extract a file from a backup. But I don't think that's something...

> I don't see the point with sync replica. You can completely choose to have an RTO of 4H by restoring a backup and an RPO of 0 because you...

> Seems like it would still be a fair bit of code and maybe not less than just writing the code to pull from the PG server directly using the...

> my only hesitation is if we need to worry about the PG version force stuff and non-regular-PG pg_control files We do handle pg-version-force with this patch (the version I...

Although `floatFormat` is right next to `blcksz` so not sure how helpful it is as a canary since we already validate `blcksz`. I guess it couldn't hurt, though.

@sfrost Here's an improvement (7225b98b1) that throws an error on archive-get when the invalid checkpoint is detected. This only works when `recovery_target_timeline != current` but `latest` is the default after...

Added new validations in 63541b22 and 960b4358 to help ensure the format of pg_control is as expected.

Not sure what `compile.sh` is, but your problem is likely that the env is different when pgbackrest is unable to load libpq.so. If you are running RHEL, why not just...

Well, as I said, there is likely something wrong with your env or libpq was not installed correctly. I don't see anything unusual about how you are building, though.