v0.9.1

Release 0.9.1

Fixed

• Fix the bug that failed to run pytorchjob with RDMA.
• Fix the bug that error dispaly gpu core resources on nodes.
• Fix the bug that add evaluator and tensorboard to pod group.

Changed

• Refact installtion.
• Modify restful-serving to http-serving of deployment services.
• Optimize the operators to omit the Completed jobs into the queue.

Added

• Support modeljob adapts helm3.
• Cron workload supports custom labels.
• Java SDK submits training job with --label.
• Add resource limits for tfjob.
• Add subpathexpr for job .

Please follow the Get started Guide to install.

v0.9.0

Release 0.9.0

Fixed

• Fix the bug arena update serving with specified kubeconfig.
• Fix the bug evaluatejob status not return.
• Fix the bug not set default shell type in arena client.
• Fix the bug of install arena when kubedl-operator existed.
• Fix the bug of mpi-operator crash.

Added

• Add command 'arena model' to support model profile/benchmark/optimize/evaluate before deploy.
• Mark 'arena evaluate' as deprecated as it has merged to 'arena model evaluate'.
• Upgrade git-sync image version to support git token.
• Upgrade arena java sdk to the latest version.
• Support execute shell with custom shell type like sh or bash.
• Support --clean-task-policy for mpijiob.
• Add arena-artifacts to adapt k8s 1.22.
• Support prometheus url token.
• Upgrade the helm version to v3.7.2 and kube client version to v1.23.0.

Please follow the Get started Guide to install.

v0.8.9

Release 0.8.9

Fixed

• Fix the bug that --gpumemory is set and set nvidia ENV NVIDIA_VISIBLE_DEVICES=void for container
• Fix the bug that skip to install cronjob crd when kubedl is existed
• Fix the bug that no write permission of /tmp in tensorflow serving container

Added

• Specify the shell type by user with --shell in training jobs
• Add command 'arena whoami' to get the current user information
• Support to setup admin users to configmap 'arena-config' in namespace arena-system
• Support enqueue before scheduled

Please follow the Get started Guide to install.

v0.8.8

Release 0.8.8

Added

• Support Evaluate job
• Support --mounts-on-launcher to mount pvc for launcher
• Support monitoring for tensorflow serving
• Support --shell to specify bash or sh for training jobs
• Triton support custom command

Fixed

• Fix the bug that disable nvidia ENV(NVIDIA_VISIBLE_DEVICES=all) for none gpu request job (#662)
• Fix the bug that git password has commas and content after comma has been lost (#663)

Please follow the Get started Guide to install.

v0.8.7

Release 0.8.7

Added

• Support Isolating users in namespace
• Support --label option
• Support annotations/nodeSelector/tolerations in TensorFlow serving jobs

Fixed

• Fix the bug of running 'arena top node'

Please follow the Get started Guide to install.

v0.8.6

Release 0.8.6

Added

• Support Display GPU for listing serving jobs

Fixed

• Fix the bug that python sdk not work when requestGPUMemory field is missing for serving job information
• Fix the bug that doc error for tfjob

Changed

• Rename kubedl-operator to cron-operator

Please follow the Get started Guide to install.

v0.8.5

Release 0.8.5

Added

• Add a Nvidia triton serving job
• Add arena-uninstall script to uninstall arena

Fixed

• Fix the bug that arena cannot display total gpu memory for nodes
• Fix the bug that missing the rbac for cronjobs
• Fix cmd typo in installation/binary.md

Please follow the Get started Guide to install.

v0.8.4

Release 0.8.4

Added

• Add cronTFJob

Please follow the Get started Guide to install.

v0.8.3

Update the way to calculate gpu usage of gpushare

v0.8.2

Release 0.8.2

Added

• Support cron task
• Support RDMA device plugin

Fixed

• Fix the bug of submitting training Job with private image registry