Bad forwarding from >1.14.2

[zifeo]

The upgrade from 1.14.2 to 1.14.7 does not handle statefulset as expected.

1.14.2

INFO[03:21:36] Port-Forward: db-replicaset-client:27017 to pod db-replicaset-0:27017 
INFO[03:21:36] Port-Forward: db-replicaset-0.db-replicaset-client:27017 to pod db-replicaset-0:27017 
INFO[03:21:36] Port-Forward: db-replicaset-1.db-replicaset-client:27017 to pod db-replicaset-1:27017 
INFO[03:21:36] Port-Forward: db-replicaset-2.db-replicaset-client:27017 to pod db-replicaset-2:27017 
INFO[03:21:36] Port-Forward: db-replicaset:27017 to pod db-replicaset-0:27017 
INFO[03:21:36] Port-Forward: db-replicaset-0.db-replicaset:27017 to pod db-replicaset-0:27017 
INFO[03:21:36] Port-Forward: db-replicaset-1.db-replicaset:27017 to pod db-replicaset-1:27017 
INFO[03:21:36] Port-Forward: db-replicaset-2.db-replicaset:27017 to pod db-replicaset-2:27017 

1.14.7

INFO[03:20:34] Port-Forward: db-replicaset-client:27017 to pod db-replicaset-0:27017 
INFO[03:20:34] Port-Forward: db-replicaset-client:27017 to pod db-replicaset-0:27017 
INFO[03:20:34] Port-Forward: db-replicaset-client:27017 to pod db-replicaset-1:27017 
INFO[03:20:34] Port-Forward: db-replicaset-client:27017 to pod db-replicaset-2:27017 
INFO[03:20:34] Port-Forward: db-replicaset:27017 to pod db-replicaset-0:27017 
INFO[03:20:34] Port-Forward: db-replicaset:27017 to pod db-replicaset-0:27017 
INFO[03:20:34] Port-Forward: db-replicaset:27017 to pod db-replicaset-1:27017 
INFO[03:20:34] Port-Forward: db-replicaset:27017 to pod db-replicaset-2:27017 

Brew rollback wget https://raw.githubusercontent.com/txn2/homebrew-tap/fb0bc557ada8de65a1577565166e006ef76ff3c8/Formula/kubefwd.rb && brew reinstall kubefwd.rb && rm kubefwd.rb (brew pin kubefwd when upgrading, do not forget to unpin later).

Relates to #133 & #134.

[cjimti]

Please try the new 1.17.0. Thanks!

[zifeo]

@cjimti thanks, while the log seems fine the /etc/hosts seems not corrected (the pods get listed twice):

127.1.27.3       db-replicaset-0.db-replicaset-0.db-replicaset

[cjimti]

I am sorry to see that. I have not seen this in my testing. Can you post the service configuration, your OS, etc? I would like to try to reproduce this.

[zifeo]

@cjimti MacOS (latest) and the service is configured as follows on GKE 1.17:

kind: Service
metadata:
  annotations:
    meta.helm.sh/release-name: db
    meta.helm.sh/release-namespace: db-namespace
    service.alpha.kubernetes.io/tolerate-unready-endpoints: "true"
  creationTimestamp: "2020-02-16T15:48:39Z"
  labels:
    app: db-replicaset
    app.kubernetes.io/managed-by: Helm
    chart: db-replicaset-2.15.1
    heritage: Helm
    release: XXX
  name: db-replicaset
  namespace: db-namespace
  resourceVersion: "1765653605"
  selfLink: /api/v1/namespaces/mongo/services/db-replicaset
  uid: 5433202c-3202-11e9-8f75-48f75a3202a5
spec:
  clusterIP: None
  ports:
  - name: db
    port: 2017
    protocol: TCP
    targetPort: 2017
  publishNotReadyAddresses: true
  selector:
    app: db-replicaset
    release: XXX
  sessionAffinity: None
  type: ClusterIP

[zifeo]

@cjimti should the issue be re-opened?

[cjimti]

I believe I found the issue and fixed it in 1.17.2

[zifeo]

@cjimti I confirm 👍 thanks