On Thu, Jan 04, 2024 at 10:56:58AM -0800, Andres Freund wrote:
> But perhaps this could actually serve as a good first case for Michael's
> failure injection patch? A failure injection to make
> FileFallocate()/FileZero() fail should just be a few lines.
>
> We generally have pretty much no coverage for out-of-space, partial
> read/write, EINTR of file operations, because that's hard to do with the
> current test infrastructure. So this might be a nice first case.
Indeed. This one should be fairly simple with a single session, a
couple of SQLs and an ERROR callback.
--
Michael