After some investigation it turns out that sending follow=true in the query of any request to the Kubernetes API changes the behaviour of the response significantly. So it's critical that it is not sent if the request is not intended to stream the result. I've modified the code so we can easily distinguish between a "dump" of the logs and a stream of the logs
After some investigation it turns out that sending
follow=true
in the query of any request to the Kubernetes API changes the behaviour of the response significantly. So it's critical that it is not sent if the request is not intended to stream the result. I've modified the code so we can easily distinguish between a "dump" of the logs and a stream of the logs