NVFlare 2.1 Release

Please see the document for more details.

What's Changed

• Fix create_analytic_dxo method calling by @YuanTingHsieh in #141
• Fix poc permission by @IsaacYangSLA in #143
• Add protection to prevent loop stream logging. by @yhwen in #146
• Remove the log streaming codes. by @yhwen in #159
• Enhanced the client_runner to handle the abort command, return Return… by @yhwen in #161
• Enhanced the DXO data and meta default to empty dict. by @yhwen in #169
• added hello-pt-tb example with Learner API and tb streaming by @SYangster in #166
• Fix issue 156 on numpy default casting by @IsaacYangSLA in #172
• Coding style by @IsaacYangSLA in #171
• add aggregation helper class by @holgerroth in #165
• Add to the documentation to clarify about specific cases by @nvkevlu in #175
• Fix private/fed/server docstrings by @YuanTingHsieh in #174
• Improve docstrings and coding style for fuel by @nvkevlu in #176
• Private docstring fixes by @yhwen in #181
• Standardize error messages by @SYangster in #178
• Update admin client command line interface to based on AdminAPI by @YuanTingHsieh in #162
• Fixe cross_validation locate_model should ony return the model weight… by @yhwen in #186
• Fix CredentialType by @YuanTingHsieh in #188
• improve docstrings for apis and widgets by @SYangster in #187
• Fix show_stats command by @YuanTingHsieh in #191
• Fix all issues of isort/black after including nvflare/private folder by @IsaacYangSLA in #154
• Add default FL app validator by @YuanTingHsieh in #179
• Fix admin shutdown client issue by @YuanTingHsieh in #194
• Add information in documentation to help explain fed events by @nvkevlu in #195
• Fix full model shareable generator by @YuanTingHsieh in #196
• Quit admin after timeout dev by @SYangster in #244
• HA: overseer/overseer_agent/overseer_agent_app/overseer_agent_gui by @IsaacYangSLA in #246
• Ha support by @yhwen in #261
• add dummy overseer agent by @SYangster in #262
• adding filesystem and s3 storage and persistor implementations by @SYangster in #255
• updated cli.py to enable HA. by @yhwen in #263
• remove old nfs_storage (filesystem_storage works for nfs) by @SYangster in #264
• Dev 2.1 cli change by @yhwen in #265
• Support HA in POC. by @yhwen in #270
• Add initialize by @IsaacYangSLA in #267
• Add json files and update templates by @IsaacYangSLA in #271
• Fix poc client start by @IsaacYangSLA in #272
• get open port for client_executor. by @yhwen in #275
• Minor changes to master-template.yml to support older versions of bash by @kkersten in #278
• Update with fed_admin config by @nvkevlu in #274
• Change fl_admin launch script to use fed_admin.json config for poc mode by @nvkevlu in #283
• update fed_admin_HA.json in poc to not use SSL by @nvkevlu in #284
• Update dev by @YuanTingHsieh in #285
• Fix intime accumulated aggregators and add tests by @YuanTingHsieh in #288
• Change log levels for some logs in private by @YuanTingHsieh in #289
• Fix license issue by @YuanTingHsieh in #287
• support SAG HA. by @yhwen in #290
• Bugfix HE aggregator by @holgerroth in #292
• Fix isort/black/flake8 issues and CI/CD by @YuanTingHsieh in #296
• Add props to SP dataclass by @IsaacYangSLA in #304
• Persist wf_index in server_runner by @YuanTingHsieh in #305
• make task name configurable by @holgerroth in #254
• Refactor overseer_agent to AdminAPI from cli, add ha commands by @nvkevlu in #302
• Enable poc command in 'pip install -e .' environment. by @IsaacYangSLA in #318
• support data kind WEIGHTS in HE by @holgerroth in #326
• Fix set_run_number for snapshot restore. by @yhwen in #312
• Clean up communicator by @YuanTingHsieh in #315
• Re-organise widgets and handlers by @YuanTingHsieh in #309
• Fix FLComponent docstring typo by @YuanTingHsieh in #329
• Add event_type to AnalyticsSender by @YuanTingHsieh in #330
• disable byoc in provisioning by @holgerroth in #327
• Clean up codes in nvflare/private/fed/app by @YuanTingHsieh in #286
• add storage tests by @SYangster in #325
• Add persist/restore to cyclic workflow by @IsaacYangSLA in #331
• Initial checkin of code with job management and admin commands by @nvkevlu in #332
• 343 fixes svt_privacy by @wyli in #348
• Improve fed_(server|client).json readability by @IsaacYangSLA in #347
• Storage improvements by @SYangster in #341
• move job def related into apis, and add get_apps by @nvkevlu in #350
• Fix persistable docstring by @YuanTingHsieh in #351
• Remove pickle dump/load calls from handling signature file by @IsaacYangSLA in #361
• Show warning after running poc command by @IsaacYangSLA in #363
• Rename study to project in all files in lighter by @IsaacYangSLA in #365
• Runner process by @yhwen in #369
• Make study spec consistent with study tool by @IsaacYangSLA in #359
• study manager by @IsaacYangSLA in #368
• Add study config user app by @IsaacYangSLA in #345
• Requires implementation of fire_event method for ServerEngineSpec by @YuanTingHsieh in #352
• Fix import issues by @IsaacYangSLA in #373
• Yaml loader known to unsafe. Switch to yaml's safe_loader to reduce … by @IsaacYangSLA in #380
• Fix some uncaught yaml loader codes. Replace them with safe_load by @IsaacYangSLA in #383
• Remove refresh api as it wipes out all records. by @IsaacYangSLA in #385
• Fix a few issues on study manager by @IsaacYangSLA in #375
• removed pickle from storage by @SYangster in #381
• Remove pickle from job commands by @nvkevlu in #382
• Update docstrings by @YuanTingHsieh in #390
• Multi run support by @yhwen in #388
• add to docstrings and add method to get app content from Job by @nvkevlu in #394
• Fix job def and manager specs by @YuanTingHsieh in #396
• fix upload and download job commands, add clone_job by @nvkevlu in #397
• fixes the issue with keys in meta not being str by @nvkevlu in #399
• Add Job scheduler by @YuanTingHsieh in #400
• update the commands in the FLAdminAPI by @nvkevlu in #401
• Fix gunicorn cert issues by @IsaacYangSLA in #402
• Fix docstrings by @YuanTingHsieh in #408
• The privilege yaml file must be signed and loaded by secure content s… by @IsaacYangSLA in #403
• Add system state into overseer reply. by @IsaacYangSLA in #410
• Re-factor unit test's folder by @YuanTingHsieh in #405
• Scheduler integration by @yhwen in #411
• Added init.py, made the ABORT command invisible. by @yhwen in #413
• Added components for poc mode. by @yhwen in #414
• Enable components in fed_server.json by @IsaacYangSLA in #404
• keep running clients for job. by @yhwen in #417
• Fix exception messages by @YuanTingHsieh in #415
• Fix store in study manager as it's now a component by @IsaacYangSLA in #418
• Replace pickle in state persistence in provision cert with json by @IsaacYangSLA in #412
• Fix error when running provision with different overseer by @IsaacYangSLA in #409
• Added snapshot write lock. by @yhwen in #419
• Update CI/CD requirements and tests by @YuanTingHsieh in #376
• Add a new entry in fed_server.json by @IsaacYangSLA in #420
• Add coverage report to unit test by @YuanTingHsieh in #421
• Fixed the cross_site_validation for multi_run. by @yhwen in #422
• Add required section for server to start normally in provision (secure) mode by @YuanTingHsieh in #424
• Add default env var PYTHONPATH if it is not set by @YuanTingHsieh in #426
• Enhance error message reporting by @YuanTingHsieh in #425
• Admin command jobs for multi-jobs by @nvidianz in #431
• Update hello-numpy-sag from app to job structure and update docs by @YuanTingHsieh in #430
• Fixed relative path problem with submit_job by @nvidianz in #433
• Update integration test cases by @YuanTingHsieh in #429
• FLARE-128: Allow missing/empty study_name by @nvidianz in #435
• Update commands by @nvkevlu in #432
• Clean up fuel/hci/server/login.py by @YuanTingHsieh in #441
• Fix typos by @YuanTingHsieh in #439
• FLARE-136: Fixed the exception in list_jobs when study is None by @nvidianz in #445
• Tests run with PASSED by @IsaacYangSLA in #427
• Remove result processor by @YuanTingHsieh in #440
• Clean up job scheduler, resource manager, and resource consumers by @YuanTingHsieh in #436
• Remove tee in sub_start.sh by @IsaacYangSLA in #447
• Add to FLAdminAPI, fix FLAdminAPI runner by @nvkevlu in #443
• Add Ditto helper to app_common, update prostate example learners by @ZiyueXu77 in #437
• fix issue of admin API not having poc_key in poc mode by @nvkevlu in #449
• clean up experience with admin client for poc mode by @nvkevlu in #450
• Move job scheduler from private to app_common by @YuanTingHsieh in #451
• Log separation by @yhwen in #446
• Enhance Job_runner logging. by @yhwen in #453
• Not allow shutdown while still jobs running. by @yhwen in #454
• Fixed codestyle. by @yhwen in #457
• Enable read the docs build by @IsaacYangSLA in #458
• Fix integration tests by @YuanTingHsieh in #455
• Add a white space to log info when deploying an application by @holgerroth in #459
• update FLAdminAPI runner by @nvkevlu in #456
• Add back HA test cases by @YuanTingHsieh in #460
• Update storage and tests by @YuanTingHsieh in #448
• Fixed the missing min_sites and required_sites for Job. by @yhwen in #461
• Generate jws-like compact serialized study info by @IsaacYangSLA in #444
• Update cifar10 data setup procedure in integration tests apps by @YuanTingHsieh in #464
• Remove wrong params local_logging by @YuanTingHsieh in #465
• Wait for admin login by @holgerroth in #463
• Changed the job scheduler to use event. by @yhwen in #468
• Only persist the FLComponent snapshot. by @yhwen in #469
• Added SecurityContentService for runner_process. by @yhwen in #473
• Add flask and gunicorn by @IsaacYangSLA in #471
• Add PyJWT license by @IsaacYangSLA in #481
• Update study def and study manager to fit study cli and all required … by @IsaacYangSLA in #480
• Clean up server deployer by @YuanTingHsieh in #479
• Clean up server engine internal spec by @YuanTingHsieh in #478
• Clean up training cmds by @YuanTingHsieh in #477
• Clean up server runner by @YuanTingHsieh in #476
• Clean up fed server by @YuanTingHsieh in #475
• add HA specific commands to FLAdminAPI by @nvkevlu in #474
• Add junitxml to unit test and update exclude var test by @YuanTingHsieh in #483
• Update job scheduler and add unit tests by @YuanTingHsieh in #472
• Update install requires by @IsaacYangSLA in #482
• Update setup.py for release by @IsaacYangSLA in #423
• Support abs path and sort list_jobs result by @nvidianz in #467
• To support Tensor used in the model weights for cross_validation. by @yhwen in #486
• Add unittests for lighter project class by @IsaacYangSLA in #488
• sync clients in runner_proess. by @yhwen in #487
• FLARE-197: Fixed the problem with empty deploy_map by @nvidianz in #494
• Add unittest cases on lighter Participant class by @IsaacYangSLA in #491
• Ensure the job exit for abort_job. by @yhwen in #496
• Added a missing check in. by @yhwen in #499
• Update dummy oa by @IsaacYangSLA in #500
• Update fed analysis example to 2.1 by @holgerroth in #495
• cifar10 example update for 2.1 by @holgerroth in #497
• Fix the list_sp index issue by @IsaacYangSLA in #503
• Fix DummyOverseerAgent overseer_info and sp_list by @IsaacYangSLA in #504
• Added job FAILED_TO_RUN status. Changed the multi-run run_number format. by @yhwen in #492
• use makedirs to auto make intermediate folders if needed by @YuanTingHsieh in #505
• Enchanced the file storage write process. by @yhwen in #502
• Fix list resource manager and add tests by @YuanTingHsieh in #485
• Modify existing premerge CI to runs-on temp-ci label by @pxLi in #507
• Enhance filesystem_storage file writting. by @yhwen in #509
• CIFAR-10 example: update readme and utility scripts by @holgerroth in #511
• Move example np codes into app_common by @YuanTingHsieh in #510
• Restructure integration tests by @YuanTingHsieh in #470
• Update 2.1 docs by @nvkevlu in #498
• Add hello examples into CI integration tests by @YuanTingHsieh in #513
• Init blossom-ci workflow by @pxLi in #517
• Fix event recorder by @YuanTingHsieh in #515
• Fix np codes by @YuanTingHsieh in #514
• update provisioning helper ui to work for 2.1.0 release by @nvkevlu in #522
• Clean up CI integration tests by @YuanTingHsieh in #516
• Blossom test by @IsaacYangSLA in #523
• Rewrote job validation and added test cases by @nvidianz in #520
• Updatedd the abort_task command usage. by @yhwen in #489
• pass the SP target info to the client process. by @yhwen in #528
• Add license files for tools needed for prostate by @ZiyueXu77 in #525
• Update readme by @YuanTingHsieh in #526
• Fix client not online deploy error by @yhwen in #529
• Removed studies from submit_job and list_jobs by @nvidianz in #532
• Removed study_name from meta.json for all examples by @nvidianz in #534
• address issues from QA by adding to docs, removing delete_job by @nvkevlu in #530
• remove study from docs and provisioning UI to match project.yml updates by @nvkevlu in #536
• Change the default value of ignore_result_error in SAG workflows by @YuanTingHsieh in #535
• Remove Study-related info from lighter/apis/poc/setup.py/requirements… by @IsaacYangSLA in #533
• Snapshot concurrent persist by @yhwen in #512
• Heartbeat jobs by @yhwen in #524
• Changed to use ClientEngineSpec for type hint. by @yhwen in #537
• Added mock package. by @yhwen in #541
• Add shutdown system admin api by @IsaacYangSLA in #543
• Remove study manager from cifar10 example by @holgerroth in #542
• Add additional datasets with preprocess and testing for prostate example by @ZiyueXu77 in #452
• Add a helper function to overseer agent by @IsaacYangSLA in #544
• cifar-10: update poc multi-task instructions by @holgerroth in #550
• add validation for shutdown_system by @nvkevlu in #548
• Enable Overseer shutdown system command. by @yhwen in #547
• Fixed the issue abort_job command could not abort all running jobs. by @yhwen in #549
• Update HE exception message by @holgerroth in #552
• Update integration tests for HA by @YuanTingHsieh in #540
• CIFAR-10: use exact match to find run by @holgerroth in #551
• Update plot_tensorboard_events.py by @ZiyueXu77 in #554
• Change tmp path under tmp/nvflare by @YuanTingHsieh in #531
• Set dummy info in both init and initialize methods (to overcome a… by @IsaacYangSLA in #557
• Limit # of servers to 2 at most by @IsaacYangSLA in #555
• Remove minio server in test scripts and fix format issues by @YuanTingHsieh in #556
• Fixed the missing client after running the system for a while. by @yhwen in #559
• Update supervised_learner.py by @ZiyueXu77 in #560
• Add overseer integration tests by @IsaacYangSLA in #493
• logout from admin api after submitting job by @holgerroth in #565
• Update plot_tensorboard_events.py by @ZiyueXu77 in #564
• Remove minio by @YuanTingHsieh in #568
• add check to throw exception if Python version is >= 3.9 by @nvkevlu in #567
• Sort the jobs in scheduler based on submit time by @YuanTingHsieh in #566
• remove study store in cifar10 example project by @holgerroth in #569
• Job validator addition checks by @chesterxgchen in #546
• Update hello-pt and hello-pt-tb example by @YuanTingHsieh in #561
• fix issue with FLAdminAPI stalling by @nvkevlu in #572
• Add execution exception status to a job by @YuanTingHsieh in #558
• Remove reference implemenation of sql/redis stores from nvflare.ha.ov… by @IsaacYangSLA in #573
• Remove S3 storage unit test by @YuanTingHsieh in #574
• Add set -e to integration test script by @YuanTingHsieh in #577
• runtests scripts refactoring by @chesterxgchen in #571
• Split snapshot file by @yhwen in #582
• Update project.yml by @IsaacYangSLA in #585
• update hello-monai README.md by @chesterxgchen in #584
• added 30 seconds delay to terminate the child process. by @yhwen in #579
• Fix handling of config exception during job running by @YuanTingHsieh in #580
• revise wording of validation for show_stats by @nvkevlu in #588
• add tests for running through FLAdminAPI commands by @nvkevlu in #581
• Fix snapshot persistor with storage by @YuanTingHsieh in #589
• hide download commands with issues that will be worked on in next rel… by @nvkevlu in #590
• Added authz for submit_job and list_jobs by @nvidianz in #587
• Init pre-merge scripts by @YanxuanLiu in #527
• Update cifar10 project file by @holgerroth in #586
• avoid swallow exception by @chesterxgchen in #591
• Fixed the snapshot log path info. by @yhwen in #592
• Remove restriction of create inside preexisting in storage by @YuanTingHsieh in #595
• Add cancelled resources to front by @YuanTingHsieh in #594
• Update jobs storage location in cifar-10 project by @holgerroth in #598
• Use PCI_BUS_ID as CUDA_DEVICE_ORDER in GPUResourceConsumer by @YuanTingHsieh in #596
• Support @ALL in site lists of deploy_map by @nvidianz in #599
• fixed the aux communication error under load. by @yhwen in #603
• dev-2.1 docs restructure by @kkersten in #601
• Support @ALL in authz_func by @nvidianz in #606
• hide command scope on CLI interface to reduce confusion by @nvkevlu in #605
• Fixed the abort_job random EOFError. by @yhwen in #597
• Fix job schedule and deploy with @ALL by @YuanTingHsieh in #607
• cherrypick the monitor parent process exit to dev-2.1. by @yhwen in #610
• Fix resource manager by @YuanTingHsieh in #604
• fix links, add information to docs raised from QA by @nvkevlu in #609
• Added proper system shutdown to avoid random Bad file descriptor errors. by @yhwen in #611
• update provisioning ui to generate updated project.yml by @nvkevlu in #593
• Created test-cases for authorization by @nvidianz in #615
• fix shutdown_system command by @nvkevlu in #612
• Fix resource manager clean up thread calling by @YuanTingHsieh in #613
• add HA integration tests with 2 servers by @nvkevlu in #616
• Fix check client replies by @YuanTingHsieh in #620
• Corrected authorization behavior for abort_job and added test cases by @nvidianz in #622
• cifar-10: don't set cuda visible devices by @holgerroth in #617
• Add specific versions to requirements-min.txt by @YuanTingHsieh in #621
• Fix abort job when server execution exception by @YuanTingHsieh in #623
• fixed jobs run race condition issue. by @yhwen in #624
• Treat several messages as NON-ERROR execution by @YuanTingHsieh in #627
• Refactor Brats to 2.1 by @ZiyueXu77 in #563
• Fixed the run_number type hint. by @yhwen in #632
• Fix a situation where server start success but client start failed by @YuanTingHsieh in #630
• fed_analysis: support new dataset with added masks by @holgerroth in #633
• make additions in docs to address various issues by @nvkevlu in #635
• Fix admin logout issue when overseer is dead by @IsaacYangSLA in #628
• Enhanced the child_process logging. by @yhwen in #637
• Enhance the abort_train with retry. by @yhwen in #634
• Add missing import by @YuanTingHsieh in #639
• Fix unit test run number issue by @YuanTingHsieh in #640
• dev-2.1 docs update - overview, examples, programming guide by @kkersten in #642
• Update job meta validator test by @YuanTingHsieh in #644
• cifar-10: Use unique dir for datasplits by @holgerroth in #643
• Fix issue #638 by removing DistributionBuilder by @IsaacYangSLA in #645
• Deprecate accumulate weighted aggregator by @YuanTingHsieh in #647
• Remove fl_ctx in init method of job meta validator by @YuanTingHsieh in #646
• update links in the examples to the location for 2.1.0 docs by @nvkevlu in #641
• Show better messages to users when starting overseer in POC mode by @IsaacYangSLA in #650
• Enable per participant component specification by @IsaacYangSLA in #651
• Replace run_number with job_id in most occurrences by @nvkevlu in #654
• Added support for name in meta.json by @nvidianz in #653
• Fix hello examples by @YuanTingHsieh in #656
• Add new test case to security unit test by @YuanTingHsieh in #652
• Clean up secure content check method by @YuanTingHsieh in #648
• Remove overseer from poc by @IsaacYangSLA in #658
• Save job workspace by @yhwen in #655
• import mock from unittest by @YuanTingHsieh in #661
• Added back delete_job command, removed the delete_workspace command. by @yhwen in #662
• update run_number to job_id in examples by @nvkevlu in #657
• CIFAR-10: add download job instructions and plot downloaded results by @holgerroth in #667
• update cifar-10 project file by @holgerroth in #666
• fed_analysis example: use download_job command; don't specify cuda devices by @holgerroth in #669
• Fix integration tests by @YuanTingHsieh in #663
• Add download_job_url to project.yml and template by @IsaacYangSLA in #671
• Delete job command enhance by @yhwen in #670
• Improve response on poc command by @IsaacYangSLA in #664
• make commands consistent with updates for FLAdminAPI and docs by @nvkevlu in #665
• update links for 2.1.1 by @nvkevlu in #668
• Use custom StorageException instead of RuntimeError by @YuanTingHsieh in #672
• Remove some unused rights from master template yaml file by @IsaacYangSLA in #675
• Add unit test for fuel/hci/zip_utils.py by @YuanTingHsieh in #659
• Readme updates on brats and prostate by @ZiyueXu77 in #673
• Unregister client itself when client shutdown by stop_fl.sh by @IsaacYangSLA in #674
• Revert stop_fl.sh PR and add warnings when users run stop_fl.sh by @IsaacYangSLA in #679
• update provisioning UI by @nvkevlu in #678
• Update example access result part and update documentation by @YuanTingHsieh in #681
• Added run_duration for the job. by @yhwen in #680
• Removal of download_folder and env and security fix for ls by @nvidianz in #682
• Remove env command in doc by @YuanTingHsieh in #683

New Contributors

• @kkersten made their first contribution in #278
• @wyli made their first contribution in #348
• @chesterxgchen made their first contribution in #546
• @YanxuanLiu made their first contribution in #527

Full Changelog: 2.0.6...2.1.1