Re: Serverside SNI support in libpq

On Fri, May 24, 2024 at 12:55 PM Cary Huang <cary.huang@highgo.ca> wrote:
> pg_hosts should also have sslpassword_command just like in the postgresql.conf in
> case the sslkey for a particular host is encrypted with a different password.

Good point. There is also the HBA-related handling of client
certificate settings (such as pg_ident)...

I really dislike that these things are governed by various different
files, but I also feel like I'm opening up a huge can of worms by
requesting nestable configurations.

> +       if (ssl_snimode != SSL_SNIMODE_OFF)
> +               SSL_CTX_set_tlsext_servername_callback(context, sni_servername_cb);
>
> If libpq client does not provide SNI, this callback will not be called, so there
> is not a chance to check for a hostname match from pg_hosts, swap the TLS CONTEXT,
> or possibly reject the connection even in strict mode.

I'm mistrustful of my own test setup (see previous email to the
thread), but I don't seem to be able to reproduce this. With sslsni=0
set, strict mode correctly shuts down the connection for me. Can you
share your setup?

(The behavior you describe might be a useful setting in practice, to
let DBAs roll out strict protection for new clients gracefully without
immediately blocking older ones.)

Thanks,
--Jacob