Merge branch 'support_multi_host_names' of github.com:yanchengnv/NVFl…

…are into support_multi_host_names

docs/real_world_fl/overview.rst

@@ -68,7 +68,7 @@ For advanced users, you can customize your provision with additional behavior th
     - **Zip**: To create password protected zip archives for the startup kits, see :ref:`distribution_builder`
     - **Docker-compose**: Provision to launch NVIDIA FLARE system via docker containers. You can customize the provisioning process and ask the provisioner to generate a docker-compose file. This can be found in :ref:`docker_compose`.
     - **Docker**: Provision to launch NVIDIA FLARE system via docker containers. If you just want to use docker files, see :ref:`containerized_deployment`.
-    - **Helm**: To change the provisioning tool to generate an NVIDIA FLARE Helm chart for Kubernetes deployment, see :ref:` helm_chart`.
+    - **Helm**: To change the provisioning tool to generate an NVIDIA FLARE Helm chart for Kubernetes deployment, see :ref:`helm_chart`.
     - **CUSTOM**: you can build custom builders specific to your needs like in :ref:`distribution_builder`.
 
 Package distribution

examples/advanced/vertical_federated_learning/cifar10-splitnn/requirements.txt

@@ -1,6 +1,6 @@
-nvflare~=2.5.0rc
+nvflare~=2.5.0
 torch
 torchvision
 tensorboard
-openmined.psi==1.1.1
+openmined.psi==2.0.5
 pandas

nvflare/app_opt/pt/job_config/model.py

@@ -24,7 +24,7 @@
 
 
 class PTModel:
-    def __init__(self, model, persistor: Optional[ModelPersistor], locator: Optional[ModelLocator]):
+    def __init__(self, model, persistor: Optional[ModelPersistor] = None, locator: Optional[ModelLocator] = None):
         """PyTorch model wrapper.
 
         If model is an nn.Module, add a PTFileModelPersistor with the model and a TFModelPersistor.

nvflare/private/fed/app/simulator/simulator_runner.py

@@ -450,8 +450,6 @@ def simulator_run_main(self):
             try:
                 self.create_clients()
                 self.server.engine.run_processes[SimulatorConstants.JOB_NAME] = {
-                    RunProcessKey.LISTEN_PORT: None,
-                    RunProcessKey.CONNECTION: None,
                     RunProcessKey.CHILD_PROCESS: None,
                     RunProcessKey.JOB_ID: SimulatorConstants.JOB_NAME,
                     RunProcessKey.PARTICIPANTS: self.server.engine.client_manager.clients,

nvflare/private/fed/client/client_engine.py

@@ -24,7 +24,6 @@
 from nvflare.apis.fl_constant import FLContextKey, MachineStatus, SystemComponents, WorkspaceConstants
 from nvflare.apis.fl_context import FLContext, FLContextManager
 from nvflare.apis.workspace import Workspace
-from nvflare.fuel.utils.network_utils import get_open_ports
 from nvflare.private.defs import ERROR_MSG_PREFIX, ClientStatusKey, EngineConstant
 from nvflare.private.event import fire_event
 from nvflare.private.fed.server.job_meta_validator import JobMetaValidator
@@ -161,15 +160,12 @@ def start_app(
 
         self.logger.info("Starting client app. rank: {}".format(self.rank))
 
-        open_port = get_open_ports(1)[0]
-
         server_config = list(self.client.servers.values())[0]
         self.client_executor.start_app(
             self.client,
             job_id,
             self.args,
             app_custom_folder,
-            open_port,
             allocated_resource,
             token,
             resource_manager,

nvflare/private/fed/client/client_executor.py

@@ -42,7 +42,6 @@ def start_app(
         job_id,
         args,
         app_custom_folder,
-        listen_port,
         allocated_resource,
         token,
         resource_manager,
@@ -56,7 +55,6 @@ def start_app(
             job_id: the job_id
             args: admin command arguments for starting the FL client training
             app_custom_folder: FL application custom folder
-            listen_port: port to listen the command.
             allocated_resource: allocated resources
             token: token from resource manager
             resource_manager: resource manager
@@ -149,7 +147,6 @@ def start_app(
         job_id,
         args,
         app_custom_folder,
-        listen_port,
         allocated_resource,
         token,
         resource_manager: ResourceManagerSpec,
@@ -163,7 +160,6 @@ def start_app(
             job_id: the job_id
             args: admin command arguments for starting the worker process
             app_custom_folder: FL application custom folder
-            listen_port: port to listen the command.
             allocated_resource: allocated resources
             token: token from resource manager
             resource_manager: resource manager
@@ -208,8 +204,6 @@ def start_app(
 
         with self.lock:
             self.run_processes[job_id] = {
-                RunProcessKey.LISTEN_PORT: listen_port,
-                RunProcessKey.CONNECTION: None,
                 RunProcessKey.CHILD_PROCESS: process,
                 RunProcessKey.STATUS: ClientStatus.STARTING,
             }

nvflare/private/fed/server/server_engine.py

@@ -50,7 +50,6 @@
 from nvflare.fuel.f3.cellnet.defs import MessageHeaderKey
 from nvflare.fuel.f3.cellnet.defs import ReturnCode as CellMsgReturnCode
 from nvflare.fuel.utils.argument_utils import parse_vars
-from nvflare.fuel.utils.network_utils import get_open_ports
 from nvflare.fuel.utils.zip_utils import zip_directory_to_bytes
 from nvflare.private.admin_defs import Message, MsgHeader
 from nvflare.private.defs import CellChannel, CellMessageHeaderKeys, RequestHeader, TrainingTopic, new_cell_message
@@ -179,13 +178,11 @@ def start_app_on_server(self, run_number: str, job: Job = None, job_clients=None
             if not isinstance(job, Job):
                 return "Must provide a job object to start the server app."
 
-            open_ports = get_open_ports(2)
             self._start_runner_process(
                 self.args,
                 app_root,
                 run_number,
                 app_custom_folder,
-                open_ports,
                 job.job_id,
                 job_clients,
                 snapshot,
@@ -233,7 +230,6 @@ def _start_runner_process(
         app_root,
         run_number,
         app_custom_folder,
-        open_ports,
         job_id,
         job_clients,
         snapshot,
@@ -244,7 +240,6 @@ def _start_runner_process(
         if app_custom_folder != "":
             add_custom_dir_to_path(app_custom_folder, new_env)
 
-        listen_port = open_ports[1]
         if snapshot:
             restore_snapshot = True
         else:
@@ -289,8 +284,6 @@ def _start_runner_process(
 
         with self.lock:
             self.run_processes[run_number] = {
-                RunProcessKey.LISTEN_PORT: listen_port,
-                RunProcessKey.CONNECTION: None,
                 RunProcessKey.CHILD_PROCESS: process,
                 RunProcessKey.JOB_ID: job_id,
                 RunProcessKey.PARTICIPANTS: job_clients,

nvflare/private/fed/simulator/simulator_client_engine.py

@@ -25,8 +25,6 @@ def __init__(self, client, args, rank=0):
         fl_ctx.set_prop(FLContextKey.SIMULATE_MODE, True, private=True, sticky=True)
 
         self.client_executor.run_processes[SimulatorConstants.JOB_NAME] = {
-            RunProcessKey.LISTEN_PORT: None,
-            RunProcessKey.CONNECTION: None,
             RunProcessKey.CHILD_PROCESS: None,
             RunProcessKey.STATUS: ClientStatus.STARTED,
         }

research/fed-bpt/README.md

@@ -14,7 +14,7 @@ The models code is copied from the [transformers](https://github.com/huggingface
 We recommend creating a [conda environment](https://www.anaconda.com) following [BBT](https://github.com/txsun1997/Black-Box-Tuning#prepare-your-environment) 
 with the addition of installing NVFlare for running federated learning and some other updates:
 ```commandline
-conda create --name fedbpt python=3.8
+conda create --name fedbpt python=3.12
 conda activate fedbpt
 pip install -r requirements.txt
 ```

research/fed-bpt/requirements.txt

@@ -1,8 +1,9 @@
 transformers==4.38.2
 fastNLP==0.6.0
 datasets
-cma
+cma==3.4.0
 scikit-learn
 tensorboard
 cvxopt
-nvflare~=2.4.1rc
+nvflare~=2.5.1rc
+