On Mon, Apr 15, 2019 at 09:07:16PM -0400, Tom Lane wrote:
> If we're failing to remove it, and it's below the desired freeze
> horizon, then we'd darn well better freeze it instead, no?
>
> Since we know that the tuple only just became dead, I suspect
> that the case would be unreachable in practice. But the approach
> you propose risks violating the invariant that all old tuples
> will either be removed or frozen.
Please note that I have added an open item for this investigation (see
"topminnow triggered assertion failure with vacuum_index_cleanup").
--
Michael