Device stuck in reconnect loop while renewing SAS token #1181
Labels
Comments
|
Bump |
|
Can you tell me a bit more about your use case?
|
|
Closed due to inactivty |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Context
Description of the issue
The iot hub client appears to be stuck in a connect/disconnect loop when reauthorizing it's SAS token. The issue occurs randomly across devices and can take anywhere from 12hours to several days before first occurence. After several minutes of connecting and disconnecting, my application is restarted and the azure client connects as expected.
Summary of events
Steps 4-9 continue until the application is stopped.
Code sample exhibiting the issue
class Client: def __init__(self, connection_str): self._connection_str = connection_str self._client: IoTHubDeviceClient = None def _statistics_increment(self, name): self.statistics[name] += 1 return self.statistics[name] async def open(self): log.debug('Azure open start') if self._client is not None: await self.close() self._client = IoTHubDeviceClient.create_from_connection_string(self._connection_str) await self._client.connect() self._twin = await self._client.get_twin() log.debug('TWIN: %s', self._twin) # set the message handler on the client self._client.on_message_received = self._message_handler self._client.on_twin_desired_properties_patch_received = self._twin_patch_listener self._client.on_method_request_received = self._device_method_listener log.debug('Azure open completed') async def close(self): if self._client is None: return log.info('Azure close') await self._client.disconnect() def is_connected(self): connected = False if self._client: connected = self._client.connected return connected async def message_send(self, payload): await self._client.send_message(payload) def run() while not self._exit_event.is_set(): try: log.debug(f"azure service loop - {azure} connected: {azure.is_connected() if azure else False}") if not azure and self._azure_enabled: azure = self._azure_factory() # Determine if we need to connect to the IoT Hub. We only try to reconnect on an interval to prevent spamming the # IoT Hub with connection attemps. poll_time_now = time.time() attempt_reconect = not azure.is_connected() and (poll_time_now >= (last_connect_poll_time + AZURE_OPEN_INTERVAL_SEC)) if attempt_reconect: last_connect_poll_time = poll_time_now log.debug("azure.open() start") await asyncio.wait_for(azure.open(), timeout=10.0) self._set_and_send_azure_connection_status(True) log.debug('azure.open() complete') last_disconnect_ts = 0 # Send messages - Coroutines are setup to swallow exceptions. Reconnect logic is handled by the azure.is_connected # function which is much more relaible and simpler than trying to propagate connectivity errors back to the main loop # from coroutines. if azure.is_connected(): for handler in self._d2c_handlers: while not handler.queue.empty(): _, payload, properties = handler.queue.get(False) task = asyncio.create_task(handler.upload(azure, payload, properties)) running_tasks.add(task) task.add_done_callback(running_tasks.discard) except ( azure_exceptions.ClientError, azure_exceptions.ConnectionDroppedError, azure_exceptions.ConnectionFailedError, azure_exceptions.CredentialError, azure_exceptions.NoConnectionError, azure_exceptions.OperationCancelled, azure_exceptions.OperationTimeout, azure_exceptions.ServiceError, ) as ex: log.exception('azure service exception') self._set_and_send_azure_connection_status(False) disconnect_ts = time.time() if azure: await asyncio.wait_for(azure.close(), timeout=10.0) azure = None await asyncio.sleep(1.0)Console log of the issue
See attached logs. The debug.log file contains DEBUG log statements for the 'azure.iot.device' logger.
azure_error_summary.log
azure_error_debug.log
The text was updated successfully, but these errors were encountered: