exec()'s timeout doesn't result in TimeoutError on busybox images

[craigwalton-dsit]

It seems that busybox has a simpler version of timeout which behaves differently.

pod = Pod("agent-env-7ycxi9fs-busybox-0", "agent")
result = await pod.exec(["sleep", "10"], timeout=1)
print(result)

This exits with exit code 143 (128 + 15 [SIGTERM]) rather than 124.

Note: the timeout does still kick in, but doesn't manifest itself as a raised TimeoutError, rather just an ExecResult with success false.

We could recognise 143 in addition to 128 as signalling a timeout error e.g.

# 124 is the exit code for the GNU Coreutils `timeout` command. Busybox's
# implementation of the `timeout` command results in 128 + 15 (SIGTERM) = 143.
if timeout is not None and result.returncode in (124, 143):

but this would incorrectly classify some commands which are killed in ways unrelated to timeout errors as TimeoutErrors. Here is an example of an agent command accidentally killing itself - because pkill -f python3 matches itself (because the command was /bin/bash -c "<the agent code which includes "python3" as a string>).

# Start the Python HTTPS server in the background
python3 -c "
import http.server
import ssl
import threading

def run_server():
    httpd = http.server.HTTPServer(('0.0.0.0', 4443), http.server.SimpleHTTPRequestHandler)
    httpd.socket = ssl.wrap_socket(httpd.socket, certfile=\\'secure.com.crt\\', keyfile=\\'secure.com.key\\', server_side=True)
    print(\\'Server running on https://0.0.0.0:4443\\')
    httpd.serve_forever()

server_thread = threading.Thread(target=run_server)
server_thread.start()
" &

# Wait for the server to start
sleep 2

# Try to connect to the server using curl
curl -k https://localhost:4443

# Stop the Python server
pkill -f "python3 -c"