* Add TFV1TrainLoopMonitorCallback.

* Remove the parameter num_workers in TFV1TrainLoopMonitorCallback.is_critical_pod

* Update comments.

* Merge two if conditions into one.

* Composite TFV1PSStrategyTrainLoopMonitorCallback if _is_tfv1_ps_strategy_training is True.

* Set the exit_code according to the success parameter in Master.request_stop.

* Add message content in request_stop.

* Organize the master exit logic.

* Resolve test failure.

* Resolve typo.

* Update the version releasing doc. (#2474)

* Update the release step: cherry-pick the fix commit from develop to release branch

* Fix typo

* Update the versioning doc.

* Add cluster_spec_json in EXCLUDE_PRINT_ARGS (#2479)

* Implement the fail fast mechanism of master. (#2480)

* Add TFV1TrainLoopMonitorCallback.

* Remove the parameter num_workers in TFV1TrainLoopMonitorCallback.is_critical_pod

* Update comments.

* Merge two if conditions into one.

* Composite TFV1PSStrategyTrainLoopMonitorCallback if _is_tfv1_ps_strategy_training is True.

* Set the exit_code according to the success parameter in Master.request_stop.

* Add message content in request_stop.

* Organize the master exit logic.

* Resolve test failure.

* Resolve typo.

* add pod status change log (#2483)

* Fix model cuda (#2484)

* Relaunch worker on failure (#2485)

* relaunch worker on failure

* only relaunch in PS strategy

* Create an ElasticImageFolder for PyTorch. (#2486)

* Develop an image folder dataset for PyTorch

* Add docstring

* Check whether to register hooks according to HOROVOD_ELASTIC (#2487)

* Check whether to register hooks according to HOROVOD_ELASTIC

* Register hooks

* Remove "elasticdl-" prefix to ps/worker pod name (#2489)

* remove prefix to ps/worker pod name

* fix tests

* fix black

Co-authored-by: brightcoder01 <55301748+brightcoder01@users.noreply.github.com>
Co-authored-by: Qinlong Wang <WangQL1201@outlook.com>

* Update the version releasing doc. (#2474)

* Update the release step: cherry-pick the fix commit from develop to release branch

* Fix typo

* Update the versioning doc.

* Add cluster_spec_json in EXCLUDE_PRINT_ARGS (#2479)

* Implement the fail fast mechanism of master. (#2480)

* Add TFV1TrainLoopMonitorCallback.

* Remove the parameter num_workers in TFV1TrainLoopMonitorCallback.is_critical_pod

* Update comments.

* Merge two if conditions into one.

* Composite TFV1PSStrategyTrainLoopMonitorCallback if _is_tfv1_ps_strategy_training is True.

* Set the exit_code according to the success parameter in Master.request_stop.

* Add message content in request_stop.

* Organize the master exit logic.

* Resolve test failure.

* Resolve typo.

* add pod status change log (#2483)

* Fix model cuda (#2484)

* Relaunch worker on failure (#2485)

* relaunch worker on failure

* only relaunch in PS strategy

* Create an ElasticImageFolder for PyTorch. (#2486)

* Develop an image folder dataset for PyTorch

* Add docstring

* Check whether to register hooks according to HOROVOD_ELASTIC (#2487)

* Check whether to register hooks according to HOROVOD_ELASTIC

* Register hooks

* Remove "elasticdl-" prefix to ps/worker pod name (#2489)

* remove prefix to ps/worker pod name

* fix tests

* fix black

* Develop an API to get training epoch (#2488)

* Check whether to register hooks according to HOROVOD_ELASTIC

* Develop an API to get training epoch

* Register hooks

* Add unittest

* Fic by comments

* Fix unittest

Co-authored-by: brightcoder01 <55301748+brightcoder01@users.noreply.github.com>
Co-authored-by: HT <tenn_2001c@yahoo.com>