Re: Bump soft open file limit (RLIMIT_NOFILE) to hard limit on startup

On Tue, 11 Feb 2025 at 22:18, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>
> Andres Freund <andres(at)anarazel(dot)de> writes:
> > My suggestion would be to redefine max_files_per_process as the number of
> > files we try to be able to open in backends. I.e. set_max_safe_fds() would
> > first count the number of already open fds (since those will largely be
> > inherited by child processes) and then check if we can open up to
> > max_files_per_process files in addition. Adjusting the RLIMIT_NOFILE if
> > necessary.
>
> Seems plausible. IIRC we also want 10 or so FDs available as "slop"
> for code that doesn't go through fd.c.

Attached is a patchset that does this. I split off the pgbench change,
which is in the first patch. The change to postmaster is in the
second. And the 3rd patch is a small follow up to make it easier to
notice that max_safe_fds is lower than intended.