GitLab

Designed new interface to support reusable training service, currently only applies to OpenPAI, and default disabled.

Replace trial_keeper.py to trial_runner.py, trial_runner holds an environment, and receives commands from nni manager to run or stop an trial, and return events to nni manager.
Add trial dispatcher, which inherits from original trianing service interface. It uses to share as many as possible code of all training service, and isolate with training services.
Add EnvironmentService interface to manage environment, including start/stop an environment, refresh status of environments.
Add command channel on both nni manager and trial runner parts, it supports different ways to pass messages between them. Current supported channels are file, web sockets. and supported commands from nni manager are start, kill trial, send new parameters; from runner are initialized(support some channel doesn't know which runner connected), trial end, stdout ((new type), including metric like before), version check (new type), gpu info (new type).
Add storage service to wrapper a storage to standard file operations, like NFS, azure storage and so on.
Partial support run multiple trials in parallel on runner side, but not supported by trial dispatcher side.
Other minor changes,

Add log_level to TS UT, so that UT can show debug level log.
Expose platform to start info.
Add RouterTrainingService to keep origianl OpenPAI training service, and support dynamic IOC binding.
Add more GPU info for future usage, including GPU mem total/free/used, gpu type.
Make some license information consistence.
Fix async/await problems on Array.forEach, this method doesn't support async actually.
Fix IT errors on download data, which causes by my #2484 .
Accelerate some run loop pattern by reducing sleep seconds.

update nni manager eslint rules.

1. add rule to ignore unused args, and treat other unused as error.
2. add run to treat no explicit function return type as error.
3. fix all errors.

To support Windows node in remote mode, this PR adds a layer of commands (osCommands) to deal difference between Windows and Unix-like OS. To share code, ShellExecutor is added to enrich original SshClient class.

I will implement windows version commands in next phase.

This pattern can be expanded to Local or other platform in future, so I moved related code to common folder for sharing.

Add shell support for ssh connection, so that remote script can be started with user environment.

Minor fixes,

1. Fix gpu_metrics_collector to support pyenv. As pyenv will create one more process, so that original pgrep code always got extra processes, and cannot start gpu_metrics_collector.
2. Fix NASUI failure on dev-install-node-modules, to create subfolder every time.
3. Fix MakeFile to reduce mis-created links, and other minor issues.
4. Add node --watch for nni_manager for better dev experience.

* Fix local system as remote machine issue #1852

* enable eslint

* remove tslint

* update eslint rules
* auto fix eslint
* manually fix eslint (#1833)

* fix gpu script permission issue

* make gpu tool local to user

* Refactor web UI to support incremental metric loading

* refactor

* Remove host job

* Move sequence ID to NNI manager

* implement incremental loading

* squash commits in v1.0 first round bug bash

* fix local and remote training services tslint

* test python

* test python36

* debug python

* debug python

* debug

* python version

* test python

* debug

* install nni

* install nni

* test powershell

* debug python

* test

* test python

* use python

* test python

* test python

* test

* update

* test powershell

* debug python

* debug python

* debug python

* debug powershell

* debug

* debug

* debug install.ps1

* add continueOnError: true

* debug

* debug

* update

* update

* add unittest

* test node

* update

* update joi

* debug joi

* add joi

* debug joi

* Update install

* update

* update

* add unittest

* add convert command

* add example

* fix windows commands

* debug

* fix tensorflow version

* fix pipeline

* update

* add gpu logic in windows

* update

* update

* debug

* fix commands

* fix commands

* update

* update

* Fix comments

* update

* fix kill command

* fix package.json

* Update package.json

* Refactor runScript

* Fix bug

* Fix comments

* Fix execKill

* Update

* Update

* Add unittest back

* Rollback install node

* Fix gpu memory

* Update

* Rollback check process

* Update mnist-hyperband.test.yml

* Update pipelines-it-local-windows.yml

* Update uninstall.ps1

* Fix virtual environment

* Fix tar

* Fix isAlive

* change gpu index logic

* test gpu index

* fix pipeline

* add cifar10

* fix cifar10

* remove gpu in cifar10

* test mnist gpu

* update

* debug

* Fix comments

* debug

* Update install.ps1

* debug

* update gpu metrics shell

* debug

* debug

* debug

* debug

* debug

* debug sigbreak

* Preinstall node-pre-gyp

* Update Installation.md

* Update Installation.md

* Remove install node-pre-gyp

* use taskkill to stop node process

* use ctl+c event to stop process

* add sigtrem signal in stop logic

* add ctl+break command

* Update isAlive

* debug sigterm

* Update pypi readme

* Update

* fix stop logic

* fix pipeline, add cifar10

* revert mnist, remove gpu

* Fix virtualenv

* Fix comments

* Update

* Update

* Fix install

* Update install.ps1

* Update install.ps1

* Fix comments

* Fix virtualenv install

* Update

* Update

* Fix comments

* Update

* Update install.ps1

* Update

* Update localTrainingService.ts

* Update

* Update

* Update

* Update

* Update

* Update util.ts

* Update utils.ts

* Fix system slash

* Update tmp dir

* Fix system slash

* Use python3 in remote

* Write tar command to file

* Update tar

* Update

* Update

* Fix stop

* Update StopSignal type

* Add removeTrialJobMetricListener

* remove Listeners

* Update listener

* Update

* Use Temp dir

* Use Temp dir

* Add remote windows pipeline

* Update pipelines-it-remote-windows.yml

* Update

* remote build wheel

* Update pipelines-it-remote-windows.yml

* debug

* debug

* Use docker source install

* Update

* Update

* Rollback remote build wheel

* Use self node and yarn

* Fix docker source install

* Rollback Makefile

* Upgrade docker pip

* Update

* Update

* Remote build wheel

* Use inline runOptions

* Hide wget output

* Add continueOnError

* Update

* Update

* Update

* Upgrade pip

* Add chmod

* Update

* debug

* Update

* Use pscp

* Update

* Download putty

* Update

* Update

* Update

* Update

* Update

* Update

* Update

* Update

* Update

* debug

* exclude metis

* Refactor pathJoin

* Update

* debug metis

* debug metis

* Update

* Update dependency

* Fix comments

* Update

* Fix tslint

* Fix comments

* Fix comments

* add doc

* Fix comments

* Update

* Update doc

If user set remoteloggingType in config file, log content will not be transmitted from trialkeeper

check nni version in trialkeeper, to make sure the version of trialkeeper is consistent with trainingService
add a debug mode in config file

SSH client has a max number of open channels for a connection, if we set the number of trialCurrency too big, our ssh client will exec command using ssh frequently, then we will meet the error of Error: (SSH) Channel open failure: open failed.
Refactor the code, set one connection has a max trial concurrency, when the number of trial reach the ssh connection restriction, will create a new ssh connection to exec trial commands.

* Fix a race condition bug that does not store Trial Job cancel status correctly

* fix remote bug

* add document

* add document

* update

* update

* update

* update

* fix remote issue

* fix forEach

* update doc according to comments

* update

* update

* update

* remove 'any more'

* add base version for remote-log

* change launcher.py

* test

* basic version

* debug

* debug

* basic work version

* fix code

* update disable_log

* remove unused line

* add diable log in kubernetesTrainingService

* add detect frameworkcontroller

* fix comment

* update

* update

* fix kubernetesData

* debug

* debug

* debug

* fix comment

* fix conflict

* remove local temp files

* revert launcher.py

* update code by comments

* remove disableLog

* remove disable Log

* set timeout for cleanup

* fix code by comments

* update variable names

* add comments

* add delay function

* update

* update

* update by comments

* add  in remote script path

* rename variables

* update variable name

* add mkdir -p for subfolder

* Pull code (#22)

* Support distributed job for frameworkcontroller (#612)

support distributed job for frameworkcontroller

* Multiphase doc (#519)

* multiPhase doc

* updates

* updates

* Add time parser for 'nnictl update duration' (#632)

Current nnictl update duration only support seconds unit, add a parser for this command to support {s, m, h, d}

* fix experiment state bug (#629)

* update top README.md (#622)

* Update README.md

* update (#634)

* Integration tests refactoring (#625)

* Integration test refactoring (#21) (#616)

* Integration test refactoring (#21)

* Refactoring integration tests

* test metrics

* update azure pipeline

* updates

* updates

* updates

* updates

* updates

* updates

* updates

* updates

* updates

* updates

* updates

* updates

* updates

* updates

* updates

* updates

* updates

* updates

* updates

* updates

* update trigger

* Integration test refactoring (#618)

* updates

* updates

* update pipeline (#619)

* update pipeline

* updates

* updates

* updates

* updates

* updates

* test pipeline (#623)

* test pipeline

* updates

* updates

* updates

* Update integration test (#624)

* Update integration test

* updates

* updates

* updates

* updates

* updates

* updates

* Revert "Pull code (#22)"

This reverts commit 62fc165ad7b2ba724eead3b99f010aa34491e2c7.

* Update nnimanager logs

* updates

* Update README.md

* Revert "Update README.md"

This reverts commit bc67061160e5d57305a6e7fb63d491d12d0e9002.

* updates

* updates

trial job could not be stopped in remote machine when experiment is stopped, because awit/async does not work normally in forEach, refer https://codeburst.io/javascript-async-await-with-foreach-b6ba62bbf404.

* Add codeDir file count validation for setClusterConfig

* fix a small bug if find command is not installed

* Remove codeDir validation for local training service

* Remove useless import