3.0.0

The minimum Echo Server version is 5.3+. From this version, you will no longer need to enable the Prometheus metrics endpoint on the server.

• Swap from Prometheus to the Usage API (#28)

2.1.0

• Added SIGTERM/SIGINT termination handler for new connections' rejection handling (#21)
• Emit Kubernetes events when updating the pod (#23)

2.0.1

• Added the getLabel() macro (6b17a31)
• Ensure to initialize the label on pods if they don't have the label set (284bd68)

2.0.0

Ensured compatibility with Echo Server 5.0.1+, since the probes API got removed. #19

The probes API served as a good way to not allow new traffic to already existent connections by failing the readiness of the pod. Since this can cause issues (and it caused), the watcher now attaches an echo.soketi.app/accepts-new-connections label (echo.soketi.app/rejects-new-connections label is the old one and it is no longer set).

The label name was changed to permit the services to monitor and attach to the given pods that will use the echo.soketi.app/accepts-new-connections labels, instead of the ones that reject connections.

To make sure the service can redirect new connections to a given pod, simply add to the service label selector the new label that network watcher will control:

spec:
  type: LoadBalancer
  ports:
    - port: 6001
      targetPort: 6001
      protocol: TCP
      name: echo
  selector:
    ...
    echo.soketi.app/accepts-new-connections: "yes"

This way, in the case that Echo Server, from a certain pod, reaches the given threshold, it will flip the label into no and the service will no longer redirect new connections to that pod.

1.1.0

• Security updates (#13, #14, #15)

1.0.5

• Extra Dockerfile cleanup (e6e2a22)

1.0.4

--no-dev on release tag

1.0.3

Updated verbosity and logic attempts.

1.0.2

Fixed in-cluster issues

1.0.1

Fixed rejection checks