In many industries (e.g. medical, energy, aeronautics) companies must produce products that can also be installed in environments where no internet connectivity is available/permitted. These same products however, are also often installed in places where they can leverage cloud (OnDemand) capabilities. If IoT-Edge ist to be applicable for such products, it must (by design) provide for:
1. Installation from download or USB-stick including configuration via config file or local API (as did Edge V1)
2. Provide and abstraction layer to the external communication (as was done via Edge V1 Modules) enabling a configurable routing to use OnPremises endpoits (REST) or OnDemand endpoints (e.g. via IoT-Hub), also via a restrictive corporate proxy (only https:443 permitted)

3. Long-term operation in the above (1. and 2.) context without the necessity of cloud connectivity.

The above 3 features wouls allow hybrid-environment product developers to use IoT-Edge as a single-source sollution.

I am sure that this scenario comprises (or will soon comprise) many thousands of product companies.

Our customer has a requirement to run multiple instances of the runtime on a single machine to isolate multiple process workflow and edge lifecycle on a single piece of hardware.

I have confirmed that this is not supported, due to the following issues:
1. iotedgectl setup will copy the –config-file to /ProgramData/azure-iot-edge overwriting the last setup call.
2. iotedgectl start will create a new docker network that is bound to the config file in ProgramData.

There is currently no workaround for RPI profile devices, but for devices that can support VMs – the runtime can be installed on a VM.

Ideally, the implementation of this follows a pattern like docker so I could run something like

iotedgectl setup edge1 –config config.json
iotedgectl start edge1
iotedgectl stop edge1
iotedgectl restart edge1
etc.

And a mapping of IoT Edge named instances to config files are stored in ProgramData.

There are two challenges that should not exist when developing custom Node.JS modules for IoT Edge:

1) Not being able to access the values in process.env without writing them programmatically into the JSON file that is passed to the instantiation of a gateway seems completely avoidable. Since, the azure-iot-gateway uses it's own Node.JS runtime, I understand why this is not happening right now but all env variables should be passed on to the individual modules themselves automatically by the azure-iot-gateway.

2) The custom modules do not easily allow for any asynchronous tasks to be performed. This is the biggest problem I having. Part of the problem here is that the individual module runtimes are still using Node.JS version 6.x.x so we are missing out on the enhanced abilities of the latest LTS Node.JS version for handing asynchronous tasks. The main issue with the edge modules I would like to see improved is the ability to perform tasks like connection to external services asynchronously within the "create" phase of a module and making sure all the connections return before the "start" phase takes place. This could be achieve be allowing the "create" phase of an edge module to return a promise. For example, this is simplified, but something like this is what I am looking to return during the "create" phase:

return createContainer()
.then((res) => true)
.catch((err) => false)

The inability to do the following is forcing me to handle all my asynchronous connection tasks in the main index.js file before the gateway starts which leads to poor programming standards because connecting to those services logically should not belong in the index.js file.

Overall, I am liking what I am seeing but under the current implementation and the two problems I have listed, it is difficult to split the edge modules into small individual files. Normally, I split my node modules into a structure like /modules/person/index.js, /modules/person/storage.js, etc. The use case for me today is that I need one of my modules to connect to azure storage before the module starts and given the two problems above I have no way to separate the main logic of the edge module from the logic behind azure storage and for handling the initialization of storage containers before the module starts.

I only have one example of where this needs to be improved at the current time.

When trying to configure the connection string for your device in iotedgectl and docker is not started an ambiguous "file or directory not found" error occurs. Ambiguous errors should be made specific.

The problem is that if one does not realize that docker is the issue there is no way to know what file or directory cannot be found. Please see the error below.

INFO: Using default IoT Edge configuration dir: /etc/azure-iot-edge
Traceback (most recent call last):
File "/usr/bin/iotedgectl", line 11, in <module>
sys.exit(coremain())
File "/usr/lib/python2.7/site-packages/edgectl/__init__.py", line 25, in coremain
return cli.execute_user_command()
File "/usr/lib/python2.7/site-packages/edgectl/cli/edgecli.py", line 69, in execute_user_command
if self._process_cli_args():
File "/usr/lib/python2.7/site-packages/edgectl/cli/edgecli.py", line 371, in _process_cli_args
return args.func(args)
File "/usr/lib/python2.7/site-packages/edgectl/cli/edgecli.py", line 389, in _parse_edge_command
return parse_funcs[args.subparser_name](args)
File "/usr/lib/python2.7/site-packages/edgectl/cli/edgecli.py", line 498, in _parse_setup_options
is_valid = self._parse_and_validate_user_input(args)
File "/usr/lib/python2.7/site-packages/edgectl/cli/edgecli.py", line 435, in _parse_and_validate_user_input
self.edge_config = parser.parse()
File "/usr/lib/python2.7/site-packages/edgectl/parser/edgeconfigparsercli.py", line 96, in parse
uri = EdgeHostPlatform.get_docker_uri()
File "/usr/lib/python2.7/site-packages/edgectl/host/edgehostplatform.py", line 20, in get_docker_uri
engine_os = dc.get_os_type()
File "/usr/lib/python2.7/site-packages/edgectl/host/dockerclient.py", line 99, in get_os_type
info = self._client.info()
File "/usr/lib/python2.7/site-packages/docker/client.py", line 174, in info
return self.api.info(*args, **kwargs)
File "/usr/lib/python2.7/site-packages/docker/api/daemon.py", line 88, in info
return self._result(self._get(self._url("/info")), True)
File "/usr/lib/python2.7/site-packages/docker/utils/decorators.py", line 46, in inner
return f(self, *args, **kwargs)
File "/usr/lib/python2.7/site-packages/docker/api/client.py", line 192, in _get
return self.get(url, **self._set_request_timeout(kwargs))
File "/usr/lib/python2.7/site-packages/requests/sessions.py", line 521, in get
return self.request('GET', url, **kwargs)
File "/usr/lib/python2.7/site-packages/requests/sessions.py", line 508, in request
resp = self.send(prep, **send_kwargs)
File "/usr/lib/python2.7/site-packages/requests/sessions.py", line 618, in send
r = adapter.send(request, **kwargs)
File "/usr/lib/python2.7/site-packages/requests/adapters.py", line 490, in send
raise ConnectionError(err, request=request)
requests.exceptions.ConnectionError: ('Connection aborted.', error(2, 'No such file or directory'))