On Thu, May 2, 2019 at 12:28 PM Alvaro Herrera <alvhe...@2ndquadrant.com> wrote:
> On 2019-May-02, Jeff Janes wrote: > > > > > When something is doomed to fail, we should report the failure as early > as > > feasibly detectable. > > I agree -- this check should be done before checking the database > contents. Maybe even before "Checking cluster versions". > It looks like it was designed for early checking, it just wasn't placed early enough. So changing it is pretty easy, as check_file_clone does not need to be invented, and there is no additional code duplication over what was already there. This patch moves the checking to near the beginning. It carries the --link mode checking along with it. That should be done as well, and doing it as a separate patch would make both patches uglier. Cheers, Jeff
pg_upgrade_filesystem.patch
Description: Binary data
