Re: Fwd: Re: A new look at old NFS readdir() problems?

On Thu, Jan 2, 2025 at 03:48:53PM -0500, Tom Lane wrote:
> Larry Rosenman <ler(at)lerctr(dot)org> writes:
> > @Tom Lane: This is what Rick Macklem (NFS dev on FreeBSD) has to say on
> > my issue.
>
> Thanks for reaching out to him. So if I'm reading this correctly,
> there's little point in filing a FreeBSD bug because it'll be
> dismissed as unfixable.
>
> This leaves us in rather a nasty position. Sure, we could rewrite
> rmtree() as Thomas suggested upthread, but I'm still of the opinion
> that that's smearing lipstick on a pig. rmtree() is the least of
> our worries: it doesn't need to expect that anybody else will be
> modifying the target directory, plus it can easily restart its scan
> without complicated bookkeeping. I doubt we can make such an
> assumption for all our uses of readdir(), or that it's okay to
> miss or double-process files in every one of them.
>
> I'm still of the opinion that the best thing to do is disclaim
> safety of storing a database on NFS.

It would be nice to have some details on the docs about why NFS can
cause problems so we have something to point to when people ask.

--
Bruce Momjian <bruce(at)momjian(dot)us> https://momjian.us
EDB https://enterprisedb.com

Do not let urgent matters crowd out time for investment in the future.