Issue while creating a Federation (interactive-api)⚙

[daviddelriod]

Hi there,

I've been trying to simulate my own FL experiment but I am encountering an error while creating the federation. My Federation cell looks like this:

client_id = 'api'
cert_dir = 'cert'
director_node_fqdn = 'localhost'
director_port = '50051'

federation = Federation(
    client_id=client_id,
    director_node_fqdn=director_node_fqdn,
    director_port=director_port, 
    tls=False
)

And I get this error while executing the cell:

_InactiveRpcError: <_InactiveRpcError of RPC that terminated with:
	status = StatusCode.UNAVAILABLE
	details = "failed to connect to all addresses"
	debug_error_string = "{"created":"@1677873513.770000000","description":"Failed to pick subchannel","file":"src/core/ext/filters/client_channel/client_channel.cc","file_line":4143,"referenced_errors":[{"created":"@1677873513.770000000","description":"failed to connect to all addresses","file":"src/core/ext/filters/client_channel/lb_policy/pick_first/pick_first.cc","file_line":398,"grpc_status":14}]}"
>

My envoy and director configuration is almost the same as the one of the TinyImage tutorial. I also provide my repo's link (the notebooks placed in the workspace folder are irrelevant except the federation cell).

I attach a screenshot of my anaconda terminals and how I launch them.

I am pretty sure that the problem is related with how I launch the scripts or maybe something related with terminal (although I am using a conda virtual environment called TFG-FL with the OpenFL library). Any kind of help would be appreciated. Thanks!

[psfoley]

Hi @daviddelriod - your configuration looks fine to me. Do you have any proxies set up on the system? If so, I would recommend setting export no_proxy=localhost for all of the terminals you shared above. Was the envoy able to connect to the director? There should be a status message that appears in both terminals.

[daviddelriod]

I have noticed that when I run either of the .sh files, it opens a cmd window for 3 seconds and shows the message bash fx command not found. This clearly indicates that the nodes are not connected.

I don't know why this happens because, as I have shown, I run the script inside a conda environment but it seems that the .sh files are executed in another external cmd. How should I execute the scripts? (Note that I have got the same error while trying this in the Virtual Studio powershell with the environment activated, in the anaconda prompt and in the windows cmd too).

[daviddelriod]

Issue solved! 🙌

The problem relied on the conda environment that I was using. I had the library working in visual studio but it was not installed in conda so I did conda install git pip and then pip install git+https://github.com/securefederatedai/openfl and the director started working (the envoys are struggling but for different errors😁).