API Gateway websockets don't stay alive

My websocket connection works fine until the keepalive timeout triggers on the client, and then it closes despite messages being sent successfully in both directions. I have tried with both .NET ClientWebSocket and python websockets. Does API Gateway not handle websocket pings? Are there any settings related to this? I could set the client keepalive really high but that seems wrong in the case of actual network issues.

Playing around with the python websocket, the connection closes as soon as it sends a ping:

import asyncio
import websockets

async def ws():
    print('connecting...')
    async with websockets.connect('wss://my-websocket-url',
            ping_interval = 1, ping_timeout = 5) as socket:
        print('connected')
        message = await socket.recv()
        print(f"< {message}")
        print('exiting')

asyncio.run(ws())

If no messages are received, the above code exits after 1 second with:

asyncio.streams.IncompleteReadError: 0 bytes read on a total of 2 expected bytes

The above exception was the direct cause of the following exception:

websockets.exceptions.ConnectionClosed: WebSocket connection is closed: code = 1006 (connection closed abnormally [internal]), no reason

This implies to me that AWS is responding to the ping with something weird? Interestingly, if I don't recv() at all then the connection does not fail. Why would receiving messages conflict with websocket pings?

Edited by: F. Spiral on Apr 3, 2019 12:21 PM

Edited by: F. Spiral on Apr 3, 2019 12:25 PM

Edited by: F. Spiral on Apr 3, 2019 12:29 PM