Hi,
On Wed, Mar 20, 2024 at 11:22:12PM +0100, Daniel Gustafsson wrote:
> > On 20 Mar 2024, at 22:21, Jacob Champion <jacob.champion@enterprisedb.com> wrote:
> >
> > On Wed, Mar 20, 2024 at 2:15 PM Jacob Champion
> > <jacob.champion@enterprisedb.com> wrote:
> >> I think solutions for case 1 and case 2 are necessarily at odds under
> >> the current design, if auth_delay relies on slot exhaustion to do its
> >> work effectively. Weakening that on purpose doesn't make much sense to
> >> me; if a DBA is uncomfortable with the DoS implications then I'd argue
> >> they need a different solution. (Which we could theoretically
> >> implement, but it's not my intention to sign you up for that. :D )
> >
> > The thread got quiet, and I'm nervous that I squashed it unintentionally. :/
> >
> > Is there consensus on whether the backoff is useful, even without the
> > host tracking? (Or, alternatively, is the host tracking helpful in a
> > way I'm not seeing?) Failing those, is there a way forward that could
> > make it useful in the future?
>
> I actually wrote more or less the same patch with rudimentary attacker
> fingerprinting, and after some off-list discussion decided to abandon it for
> the reasons discussed in this thread. It's unlikely to protect against the
> attackers we wan't to protect the cluster against since they won't wait for the
> delay anyways.
I have marked the patch "Returned with Feedback" now. Maybe I will get
back to this for v18, but it was clearly not ready for v17.
Michael