DNS discovery slow to notice scale downs

So I’m noticing that the round-robin DNS is slow to notice terminated/failed containers.
If you scale a 3-container service down to 2, you notice every third request fails for several minutes.

I’m assuming this is a weave issue?

It’s a bummer because it makes the DNS-based discovery (an awesome feature) unusable in any situations where services need to scale down without downtime.

Agree. This is really annoying. Makes it especially hard to move containers from host to host.