Feature/SK-707 | Add config support to flower adapter

examples/flower-client/README.rst

@@ -1,5 +1,5 @@
 Using Flower ClientApps in FEDn
-============================
+-------------------------------
 
 This example demonstrates how to run a Flower 'ClientApp' on FEDn.
 
@@ -15,28 +15,32 @@ Running the example
 See `https://fedn.readthedocs.io/en/stable/quickstart.html` for a general introduction to FEDn. 
 This example follows the same structure as the pytorch quickstart example. 
 
-Build a virtual environment (note that you might need to install the 'venv' package): 
+Install fedn:
 
 .. code-block::
 
-   bin/init_venv.sh
+   pip install fedn
 
-Activate the virtual environment:
+Clone this repository, then locate into this directory:
 
 .. code-block::
 
-   source .flower-client/bin/activate
+   git clone https://github.com/scaleoutsystems/fedn.git
+   cd fedn/examples/mnist-pytorch
 
-Make the compute package (to be uploaded to FEDn):
+Create the compute package (compress the 'client' folder):
 
 .. code-block::
 
-   tar -czvf package.tgz client
+   fedn package create --path client
+
+This should create a file 'package.tgz' in the project folder.
+
+Next, generate a seed model (the first model in the global model trail):
 
-Create the seed model (to be uploaded to FEDn):
 .. code-block::
 
-   python client/entrypoint init_seed
+   fedn run build --path client
 
 Next, you will upload the compute package and seed model to
 a FEDn network. Here you have two main options: using FEDn Studio 
@@ -53,28 +57,26 @@ In your Studio project:
 - From the "Sessions" menu, upload the compute package and seed model. 
 - Register a client and obtain the corresponding 'client.yaml'.  
 
-On your local machine / client (in the same virtual environment), start the FEDn client: 
-
-.. code-block::
-
-   CLIENT_NUMBER=0 FEDN_AUTH_SCHEME=Bearer fedn run client -in client.yaml --force-ssl --secure=True
-
+On your local machine / client, start the FEDn client: 
 
-Or, if you prefer to use Docker, build an image (this might take a long time):
 
 .. code-block::
 
-   docker build -t flower-client .
+   export FEDN_AUTH_SCHEME=Bearer
+   export FEDN_PACKAGE_EXTRACT_DIR=package
+   CLIENT_NUMBER=0 fedn run client -in client.yaml --secure=True --force-ssl
 
-Then start the client using Docker:
+
+Or, if you prefer to use Docker (this might take a long time):
 
 .. code-block::
 
    docker run \
    -v $PWD/client.yaml:/app/client.yaml \
    -e CLIENT_NUMBER=0 \
    -e FEDN_AUTH_SCHEME=Bearer \
-   flower-client run client -in client.yaml --secure=True --force-ssl
+   -e FEDN_PACKAGE_EXTRACT_DIR=package \
+   ghcr.io/scaleoutsystems/fedn/fedn:master run client -in client.yaml --secure=True --force-ssl
 
 
 If you are running FEDn in pseudo-local mode:
@@ -105,4 +107,6 @@ Then start the client (using Docker)
    -v $PWD/client.yaml:/app/client.yaml \
    --network=fedn_default \
    -e CLIENT_NUMBER=0 \
-   flower-client run client -in client.yaml
+   -e FEDN_AUTH_SCHEME=Bearer \
+   -e FEDN_PACKAGE_EXTRACT_DIR=package \
+   ghcr.io/scaleoutsystems/fedn/fedn:master run client -in client.yaml

examples/flower-client/client/entrypoint → examples/flower-client/client/entrypoint.py

@@ -1,4 +1,3 @@
-#!/usr/bin/env python
 import os
 
 import fire
@@ -38,7 +37,7 @@ def init_seed(out_path="seed.npz"):
     :type out_path: str
     """
     # This calls get_parameters in the flower client which needs to be implemented.
-    parameters_np = flwr_adapter.init_parameters(partition_id=_get_node_id())
+    parameters_np = flwr_adapter.init_parameters(partition_id=_get_node_id(), config={})
     save_parameters(out_path, parameters_np)
 
 
@@ -58,7 +57,7 @@ def train(in_model_path, out_model_path):
 
     # Train on flower client
     params, num_examples = flwr_adapter.train(
-        parameters=parameters_np, partition_id=_get_node_id()
+        parameters=parameters_np, partition_id=_get_node_id(), config={}
     )
 
     # Metadata needed for aggregation server side
@@ -86,7 +85,7 @@ def validate(in_model_path, out_json_path, data_path=None):
     """
     parameters_np = helper.load(in_model_path)
 
-    loss, accuracy = flwr_adapter.evaluate(parameters_np, partition_id=_get_node_id())
+    loss, accuracy = flwr_adapter.evaluate(parameters_np, partition_id=_get_node_id(), config={})
 
     # JSON schema
     report = {

examples/flower-client/client/fedn.yaml

@@ -1,5 +1,8 @@
+python_env: python_env.yaml
 entry_points:
+  build:
+    command: python entrypoint.py init_seed
   train:
-    command: python entrypoint train $ENTRYPOINT_OPTS
+    command: python entrypoint.py train
   validate:
-    command: python entrypoint validate $ENTRYPOINT_OPTS
+    command: python entrypoint.py validate

examples/flower-client/client/python_env.yaml

@@ -0,0 +1,12 @@
+name: flower-client
+build_dependencies:
+  - pip
+  - setuptools
+  - wheel==0.37.1
+dependencies:
+  - torch==2.2.1
+  - torchvision==0.17.1
+  - fire==0.3.1
+  - fedn==0.8.0
+  - flwr==1.8.0
+  - flwr-datasets[vision]==0.1.0

fedn/fedn/utils/flowercompat/client_app_adapter.py

@@ -19,10 +19,10 @@ class FlwrClientAppAdapter:
     def __init__(self, app: ClientApp) -> None:
         self.app = app
 
-    def init_parameters(self, partition_id: int):
+    def init_parameters(self, partition_id: int, config: dict = {}):
         # Construct a get_parameters message for the ClientApp
         message, context = self._construct_message(
-            MessageTypeLegacy.GET_PARAMETERS, [], partition_id
+            MessageTypeLegacy.GET_PARAMETERS, [], partition_id, config
         )
         # Call client app with train message
         client_return_message = self.app(message, context)
@@ -33,21 +33,21 @@ def init_parameters(self, partition_id: int):
                              client has implemented a get_parameters() function.")
         return parameters
 
-    def train(self, parameters: NDArrays, partition_id: int):
+    def train(self, parameters: NDArrays, partition_id: int, config: dict = {}):
         # Construct a train message for the ClientApp with given parameters
         message, context = self._construct_message(
-            MessageType.TRAIN, parameters, partition_id
+            MessageType.TRAIN, parameters, partition_id, config
         )
         # Call client app with train message
         client_return_message = self.app(message, context)
         # Parse return message
         params, num_examples = self._parse_train_message(client_return_message)
         return params, num_examples
 
-    def evaluate(self, parameters: NDArrays, partition_id: int):
+    def evaluate(self, parameters: NDArrays, partition_id: int, config: dict = {}):
         # Construct an evaluate message for the ClientApp with given parameters
         message, context = self._construct_message(
-            MessageType.EVALUATE, parameters, partition_id
+            MessageType.EVALUATE, parameters, partition_id, config
         )
         # Call client app with evaluate message
         client_return_message = self.app(message, context)
@@ -76,16 +76,17 @@ def _construct_message(
         message_type: MessageType,
         parameters: NDArrays,
         partition_id: int,
+        config: dict,
     ) -> Tuple[Message, Context]:
         parameters = ndarrays_to_parameters(parameters)
         if message_type == MessageType.TRAIN:
-            fit_ins: FitIns = FitIns(parameters=parameters, config={})
+            fit_ins: FitIns = FitIns(parameters=parameters, config=config)
             recordset = fitins_to_recordset(fitins=fit_ins, keep_input=False)
         if message_type == MessageType.EVALUATE:
-            ev_ins: EvaluateIns = EvaluateIns(parameters=parameters, config={})
+            ev_ins: EvaluateIns = EvaluateIns(parameters=parameters, config=config)
             recordset = evaluateins_to_recordset(evaluateins=ev_ins, keep_input=False)
         if message_type == MessageTypeLegacy.GET_PARAMETERS:
-            get_parameters_ins: GetParametersIns = GetParametersIns({})
+            get_parameters_ins: GetParametersIns = GetParametersIns(config=config)
             recordset = getparametersins_to_recordset(getparameters_ins=get_parameters_ins)
 
         metadata = Metadata(