Regarding the first issue of the duplicate calls, it looks like there aren't actually 2 separate http requests occurring. From what we can observe in the flame graph it looks like it's the same http request happening at 2 different layers in the application.

The first http.request span is from google-http-client (see component tag in span), and the second span is from http-url-connection.

It seems the google http client is using an underlying http client and the tracer is instrumenting both.

If you would prefer to see only one of these I can check in with the team and see if there's a way of disabling one of the two integrations.

As for this error you're seeing:
Cannot decode v0.4 traces payload: msgp: too few bytes left to read object
We've seen this associated with agent overload. The tracer has a timeout of 10 seconds, where it will drop the trace if the timeout is exceeded. To offset this, you could try increasing the timeout by adding the following parameter:

System property :
-Ddd.trace.agent.timeout=20
The environment variable equivalent is:
DD_TRACE_AGENT_TIMEOUT=20

Please restart the agent after doing so and let me know if this stops the error you're encountering.