Do not spend login attempts on an outage nobody caused

ihasmail runs in its own container, usually on its own host, so Stalwart being
briefly unreachable is an ordinary Tuesday. Sign-in handled it almost right:
a 401 is invalid_credentials, a timeout is 504 and anything else is 502, none
of which reads as a rejected password.

What it got wrong was the counting. RateLimiter.check() consumes an attempt
when it is called, and it is called before the upstream is contacted; reset()
only runs on success. So every try against an unreachable server burned a
credential attempt, and after ten of them the person was locked out for the
rest of the fifteen-minute window -- including after the server came back. A
thirty-second blip became a quarter-hour lockout, and the second failure was
entirely ihasmail's own doing.

A 401 is a judgement about the password and stays counted. A 502 or 504 is the
upstream failing to answer, says nothing about the credentials, and is now
refunded -- one attempt back, not the key cleared, so a run of real failures
with an outage in the middle still adds up. The old-server refusal refunds too:
those credentials were accepted.

Both guessing keys are refunded, not just the username one. Refunding only
that would not have fixed it -- ten retries still spend the per-address budget,
and behind one office NAT that budget belongs to the whole building, so a
company-wide outage would lock out the company.

Which needs a backstop, because "not counted" must not mean "unlimited": each
attempt still costs an outbound connection that may sit there until
UPSTREAM_TIMEOUT, and an outage is the one moment the endpoint is cheapest to
abuse. So there is a second ceiling per address, twenty times looser and never
refunded. A person retrying will not come near it; something hammering will.

Both messages now say the quiet part -- "This is not a problem with your
password" -- for somebody already worried they have forgotten it.

Closes #239.
This commit is contained in:
2026-09-02 14:08:23 -07:00
parent f6f6ce0b5e
commit 607afeb4ad
4 changed files with 173 additions and 2 deletions
+67
View File
@@ -0,0 +1,67 @@
import { test } from "node:test";
import assert from "node:assert/strict";
import { RateLimiter } from "./ratelimit.js";
/**
* The limiter's job is to slow down password guessing. #239 is about the
* attempts it takes for outcomes that were never a guess: ihasmail runs apart
* from Stalwart, so an upstream that refuses a connection is ordinary, and
* retrying through one used to spend the window and lock somebody out until
* after the cause had gone.
*/
test("check allows up to the limit and then refuses", () => {
const rl = new RateLimiter(3, 60_000);
assert.equal(rl.check("k"), true);
assert.equal(rl.check("k"), true);
assert.equal(rl.check("k"), true);
assert.equal(rl.check("k"), false);
});
test("refund gives back exactly one attempt", () => {
const rl = new RateLimiter(2, 60_000);
rl.check("k");
rl.check("k");
assert.equal(rl.check("k"), false, "spent");
rl.refund("k");
assert.equal(rl.check("k"), true, "one back");
assert.equal(rl.check("k"), false, "and only one");
});
test("refunding every attempt leaves the key spending nothing", () => {
// The outage case: every try refunded, so a person retrying through it is
// not locked out when the server returns.
const rl = new RateLimiter(2, 60_000);
for (let i = 0; i < 20; i++) {
assert.equal(rl.check("k"), true, `attempt ${i} allowed`);
rl.refund("k");
}
});
test("a run of real failures still adds up around a refunded one", () => {
// Refund takes one attempt back, not the key's whole history -- an outage in
// the middle of somebody guessing must not clear what they spent before it.
const rl = new RateLimiter(3, 60_000);
rl.check("k"); // a wrong password
rl.check("k"); // another
rl.check("k"); rl.refund("k"); // an outage, given back
assert.equal(rl.check("k"), true, "third real attempt");
assert.equal(rl.check("k"), false, "and now spent");
});
test("refunding a key that never spent anything is harmless", () => {
const rl = new RateLimiter(1, 60_000);
rl.refund("never-seen");
assert.equal(rl.check("never-seen"), true);
});
test("reset clears the key, refund does not", () => {
const rl = new RateLimiter(2, 60_000);
rl.check("k");
rl.check("k");
rl.refund("k");
assert.equal(rl.check("k"), true);
assert.equal(rl.check("k"), false);
rl.reset("k");
assert.equal(rl.check("k"), true, "reset is the successful-sign-in case");
});