Degraded performance in swarm during a scale up

[Yshayy]

What version of Traefik are you using?1.2.0-rc2

What is your environment & configuration?

Docker swarm mode.
Multiple traefik instances, multiple services.
Using traefik.backend.loadbalancer.swarm=false for a service (or just keeping it anything other then true).

What did you do?

Scaling up a service.

What did you expect to see?

Everything works and performs well.

What did you see instead?

The service endpoint have significant degraded performance during the new containers startup.

This issue doesn't reproduce when using traefik.backend.loadbalancer.swarm=true.
I don't have much experience with Traefik or swarm, but it looks to me that swarm's tasks don't change their state to running (from starting) until health-check is complete.
While swarm VIP is aware of the difference between state "starting" and "running" and does not pass traffic to these containers, I think that Traefik as of 1.2.0-rc2 ignores it.

[timoreimann]

@Yshayy did you resolve the issue? If so, how?