Closed xguerin closed 1 year ago
Some context below. No idea why we get another response after ~20 minutes or so.
Request:
Request{method=GET, url=https://kubernetes.docker.internal:6443/apis/streams.ibm.com/v1/namespaces/default/streamsprs?resourceVersion=1225180&watch=true, tags={}}
Request headers:
Origin: https://kubernetes.docker.internal:6443
Upgrade: websocket
Connection: Upgrade
Sec-WebSocket-Key: rbVgB6L78UPmkVxA+82Ppg==
Sec-WebSocket-Version: 13
Response:
Response{protocol=http/1.1, code=101, message=Switching Protocols, url=https://kubernetes.docker.internal:6443/apis/streams.ibm.com/v1/namespaces/default/streamsprs?resourceVersion=1225180&watch=true}
Response headers:
Upgrade: websocket
Connection: Upgrade
Sec-WebSocket-Accept: E3+/F9m256LMtWuBqeHZOmxPV/o=
Websocket opcode: 8, OPCODE_CONTROL_CLOSE
After prolonged executions with 0.3.0 (API version 4.11), I run into the following event pattern:
As a result of this, it looks like the events get resynchronized (or forwarded twice) to the listening application. This issue might be related: https://github.com/fabric8io/kubernetes-client/issues/1602. This is how it manifests itself: https://github.com/IBMStreams/OSStreams/issues/11.
I'll investigate to find the how, what and why, but I thought it would be worth to mention the issue here in case anything seems obvious.