Middlewares & multiple replicas

Hi all,
In 2.0, is it safe to assume that “stateful” (for lack of a better term) middlewares such as Maxconn, RateLimit, and Retry can’t communicate their state with other pods in a multi-replica deployment? Meaning, if there’s a Maxconn limit of 10, and there are 3 replicas in the Traefik deployment, the real observable maximum number of connections is 30?

Thanks!

You’re right.

On a side note, if you’re looking for “stateful” middleware, then you might want to look toward the EE version which provides this feature (thx to its clustered architecture)

1 Like