Firmware 3.92.0 interrupts the connection

[Guzz-T]

The new firmware 3.92.0 seems to interrupts the connection 1-2 per day. After that, no new values are retrieved. Reconnecting and then closing the connection for each operation might solve the problem.

[kbabioch]

Haven't updated yet, as I don't want to run into this issue :-).

Is this something that happens on the controller itself? Can it be reproduced? Maybe its worthwhile to report this to AIT?

[gerw]

In the original version of the issue, you wrote that the automatic time retrieval causes this problem. Do you no longer think that this is the problem?

Maybe the changes of the clock on the controller disturbs the connection? Then, also manual changes of the clock should cause this and one could try to reproduce the problem.

Do I get it correctly, that self._socket.recv locks forever when the connection interrupts? Then, a timeout seems to be the cleanest solution.

[Guzz-T]

After creating the original issue, I disabled the automatic time setting, but the problem persists.

As a workaround, I modified _with_lock_and_connect so that the connection is only established for data transfer:

    def _with_lock_and_connect(self, func, *args, **kwargs):
        with self._lock:
            try:
                ret_val = None
                with socket.create_connection((self._host, self._port)) as sock:
                    LOGGER.info("Connected to Luxtronik heat pump %s:%s", self._host, self._port)
                    ret_val = func(sock, *args, **kwargs)
            except socket.gaierror as e:
                LOGGER.error("Failed to connect to Luxtronik heat pump %s:%s. %s.",
                    self._host, self._port, f"Address-related error: {e}")
            except socket.timeout as e:
                LOGGER.error("Failed to connect to Luxtronik heat pump %s:%s. %s.",
                    self._host, self._port, f"Connection timed out: {e}")
            except ConnectionRefusedError as e:
                LOGGER.error("Failed to connect to Luxtronik heat pump %s:%s. %s.",
                    self._host, self._port, f"Connection refused: {e}")
            except OSError as e:
                LOGGER.error("Failed to connect to Luxtronik heat pump %s:%s. %s.",
                    self._host, self._port, f"OS error during connect: {e}")
            except Exception as e:
                LOGGER.error("Failed to connect to Luxtronik heat pump %s:%s. %s.",
                    self._host, self._port, f"Unknown exception: {e}")
        return ret_val

and keep getting

OS error during connect: [Errno 113] Host is unreachable.

It is too systematic for it to be a multi-thread access problem or something similar. Every 22 hours, access is not possible for 1 minute.

It may also only be related to my master-slave heat pump solution. These also communicate via LAN.

Note:

The current implementation returns False on OSError which is why no re-connect is performed:

def is_socket_closed(sock: socket.socket) -> bool:
    ...
    except Exception as err: 
        LOGGER.exception("Unexpected exception when checking if socket is closed", exc_info=err)
        return False

[kbabioch]

How do we want to proceed here? Seems like we still cannot really pin down the root cause of the issue. Is anyone brave enough to update to more recent versions of the firmware :-)?

[kbabioch]

@Guzz-T With #216 merged: Does this change anything (and/or solve it)? Should we keep this issue open?

[Guzz-T]

It just reconnects. Unfortunately, that doesn't solve the problem. However, it may be that we can't solve this problem at all. Nevertheless, I would leave the issue open until the next firmware release.