Problems creating statically configured middleware

Traefik Traefik v2
1

Migrating from 1.7 to 2.8.5 requires some additional middleware configuration.
I'm unable to create statically configured middlewares using the kubernetes ingress provider. Below is a simplified version of our configuration to replicate the error.

---
kind: ServiceAccount
apiVersion: v1
metadata:
  name: traefik-2-internal
  namespace: traefik-internal
---
kind: ClusterRole
apiVersion: rbac.authorization.k8s.io/v1
metadata:
  name: traefik-2-internal
rules:
  - apiGroups:
      - ""
    resources:
      - pods
      - services
      - endpoints
      - secrets
    verbs:
      - get
      - list
      - watch
  - apiGroups:
      - extensions
      - networking.k8s.io
    resources:
      - ingresses
      - ingressclasses
    verbs:
      - get
      - list
      - watch
  - apiGroups:
      - extensions
      - networking.k8s.io
    resources:
      - ingresses/status
    verbs:
      - update
  - apiGroups:
      - traefik.containo.us
    resources:
      - middlewares
      - middlewaretcps
      - ingressroutes
      - traefikservices
      - ingressroutetcps
      - ingressrouteudps
      - tlsoptions
      - tlsstores
      - serverstransports
    verbs:
      - get
      - list
      - watch
---
kind: ClusterRoleBinding
apiVersion: rbac.authorization.k8s.io/v1
metadata:
  name: traefik-2-internal
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: traefik-2-internal
subjects:
  - kind: ServiceAccount
    name: traefik-2-internal
    namespace: traefik-internal
---
apiVersion: v1
kind: ConfigMap
metadata:
  name: traefik-2-internal-config
  namespace: traefik-internal
data:
  traefik.yml: |
    accesslog:
      format: json
      bufferingsize: 100
    log:
      level: INFO
      format: json
    entrypoints:
      http:
        address: ":33180"
        http:
          middlewares:
            - traefik-internal-compress@kubernetescrd
        transport:
          respondingtimeouts:
            idletimeout: 902s
            readtimeout: 0s
            writetimeout: 0s
          lifecycle:
            gracetimeout: 902s
            requestacceptgracetimeout: 30s
    providers:
      kubernetesingress:
        labelselector: traefik=internal-https
        ingressendpoint:
          hostname: internalalb.elb.amazonaws.com
---
apiVersion: apps/v1
kind: DaemonSet
metadata:
  name: traefik-2-internal
  namespace: traefik-internal
  labels:
    app: traefik-2
spec:
  selector:
    matchLabels:
      name: traefik-2-internal
  minReadySeconds: 15
  updateStrategy:
    type: RollingUpdate
    rollingUpdate:
      maxUnavailable: 1
  template:
    metadata:
      labels:
        app: traefik-2
        name: traefik-2-internal
    spec:
      serviceAccountName: traefik-2-internal
      priorityClassName: high-priority
      securityContext:
        runAsUser: 1000
        fsGroup: 2000
        runAsNonRoot: true
      containers:
        - image: traefik:2.8.5
          name: traefik-2-internal
          ports:
            - name: http
              containerPort: 33180
              hostPort: 33180
          volumeMounts:
            - name: config-volume
              mountPath: /etc/traefik
      volumes:
        - name: config-volume
          configMap:
            name: traefik-2-internal-config
---
apiVersion: traefik.containo.us/v1alpha1
kind: Middleware
metadata:
  name: compress
  namespace: traefik-internal
spec:
  compress: {}

Then, for each matched ingress object, traefik prints the following error

{"entryPointName":"http","level":"error","msg":"middleware \"traefik-internal-compress@kubernetescrd\" does not exist","routerName":"http-my-fine-service-com@kubernetes",...}

With a corresponding ingress object definition:

apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: my-fine-service-ingress
  labels:
    traefik: internal-https
spec:
  rules:
    - host: my-fine-service.com
      http:
        paths:
          - path: /
            pathType: Prefix
            backend:
              service:
                name: my-fine-service
                port:
                  name: http

As far as I can see, I've correctly formatted the router definition based on k8s namespace specific crd-definitions. I've been banging my head for a day now debugging this, and can't seem to pinpoint the issue.