Empty reply from server on delays ≥ 30s

[anderius]

When using the endpoint /delay/XXX, we experience that is does not work as documented when the delay is 30 seconds or longer.
I start podinfo like this:

docker run -dp 9898:9898 stefanprodan/podinfo

A delay below 30 seconds works as expected:

 % curl http://localhost:9898/delay/29
{
  "delay": 29
}

A delay of 30 seconds (or above) gives no response

% curl http://localhost:9898/delay/30
curl: (52) Empty reply from server

I want to use podinfo for testing timeouts in the infrastructure, but this behavior makes that difficult.

[anderius]

The following test shows that it is not a timeout before I reach the container, as a request for /delay/40 takes 40 seconds, as it should. It just does not give a reply.

% time curl http://localhost:9898/delay/40              
curl: (52) Empty reply from server
curl http://localhost:9898/delay/40  0.00s user 0.01s system 0% cpu 40.028 total