new TonY release

v0.3.33

move System.exit(exitCode) out of try block so that finally block can…

Fix tensorboard port conflict

version bump (#462)

Co-authored-by: Cheng Ren <1428327+chengren311@users.noreply.github.com>

Enable port reuse option

version upgrade (#460)

Co-authored-by: Cheng Ren <1428327+chengren311@users.noreply.github.com>

Bug fix: Tony source archive extraction during parallel builds

• Fixed archive extracting issues with parallel workflows using unique naming (#455 )

Bug Fixes

fixes the bug where the retry fails, if the previous failures was due to heartbeat failure.

Miscellaneous Fixes

Changelog:

• Add github workflow to auto release to maven central on tags(#439)
• Add github action based build workflow (#438)
• Fix retry in case of untracked task failures (#436)
• Update README.md (#433)
• Add job_id env in container (#431)
• Added fix to return successful training with some failed worker jobs (#428)
• Exposing container diagnostic information for failed task to the tonyhistory files (#429)
• Fixing AM RPC port binding issue (#427)
• Support to specify application type (#426)
• Enable debugging with tony am (#425)
• Exposing container logs URL in tony portal (#424)
• Refine README.md about TonY in gcp (#423)
• Add support for CUDA 10.1, add generic CUDA script and update documen… (#419)
• Update install_gpu_cu10.sh (#417)
• Change default allocation timeout to -1 to avoid incompatible behavior. Bump version (#416)
• Add wait timeout for allocated container registration (#413)
• Add option to stop application on failure of certain job (#414)
• Handle untracked job type crash by failing the application for #405 (#406)
• TonY should be able to launch jobs using dependency graph (#403)

Identifies Killed Apps in TonY Portal

See #399 for more details

MXNet support, scalability improvements and bug fixes

• Scalability improvements to support 200+ concurrent executors.
• UX fix to print out am log url and logging
• Support MXNet
• Fix port race condition problem

Adds Sorting/Filtering to TonY Portal

See #354 for more details.
Released on Maven Central.

Made fetching final task status more robust

See #350 for more details. Released on Maven Central.