Kube Batch Versions Save

• master and worker start in order in a job #872
• use kubernetes default schedule policies #864
• v0.4 runtime panic #861
• Helm install fails #857
• "NewResource/Convert2K8sResource" behavior mismatch #851
• Phase of podgroup status looks incorrect #846
• Update API GroupName #815
• Migrate nodeorder and predicates plugins #814
• Add Configuration for predicate plugin to enable/disable predicates algorithm #802
• Consider support multi-containers pod error code handling #776
• Deserved attr is not correctly calculated in proportion plugin #729
• Keep backward compatibility for priority class #724
• Change return value of NodeOrderFn from int to float #708
• Add type Argument with some common parse function #704
• Replace NodeOrder with BestNode #699
• Support set default value to the configuration #695
• Add resource predicates for tasks #694
• Pass conformance test #589
• big PodGroup blocks scheduling issue #514

v0.4.2

Notes:

• Added Balanced Resource Priority
• Support Plugin Arguments
• Support Weight for Node order plugin

Issues:

• #636 add balanced resource priority (@DeliangFan)
• #638 Take init containers into account when getting pod resource request (@hex108)
• #639 Support Plugin Arguments (@thandayuthapani)
• #640 Support Weight for NodeOrder Plugin (@thandayuthapani)
• #642 Add event when task is scheduled (@Rajadeepan)
• #643 Return err in Allocate if any error occurs (@hex108)
• #644 Fix Typos (@TommyLike)
• #645 Order task by CreationTimestamp first, then by UID (@hex108)
• #647 In allocate, skip adding Job if its queue is not found (@hex108)
• #649 Preempt lowest priority task first (@hex108)
• #651 Return err in functions of session.go if any error occurs (@hex108)
• #652 Change run option SchedulePeriod's type to make it clear (@hex108)
• #655 Do graceful eviction using default policy (@hex108)
• #658 Address helm install error in tutorial.md (@hex108)
• #660 Fix sub exception in reclaim and preempt (@TommyLike)
• #666 Fix wrong caculation for deserved in proportion plugin (@zionwu)
• #671 Change base image to alphine to reduce image size (@hex108)
• #673 Do not create PodGroup and Job for task whose scheduler is not kube-batch (@hex108)

v0.4.1

Notes:

• Added NodeOrder Plugin
• Added Conformance Plugin
• Removed namespaceAsQueue feature
• Supported Pod without PodGroup
• Added performance metrics for scheduling

Issues:

• #632 Set schedConf to defaultSchedulerConf if failing to readSchedulerConf (@hex108)
• #631 Add helm hook crd-install to fix helm install error (@hex108)
• #584 Replaced FIFO by workqueue. (@k82cn)
• #585 Removed invalid error log. (@k82cn)
• #594 Fixed Job.Clone issue. (@k82cn)
• #600 Fixed duplicated queue in preempt action. (@k82cn)
• #596 Set default PodGroup for Pods. (@k82cn)
• #587 Adding node priority (@thandayuthapani)
• #607 Fixed flaky test. (@k82cn)
• #610 Updated log level. (@k82cn)
• #609 Added conformance plugin. (@k82cn)
• #604 Moved default implementation from pkg. (@k82cn)
• #613 Removed namespaceAsQueue. (@k82cn)
• #615 Handled minAvailable task everytime. (@k82cn)
• #603 Added PriorityClass to PodGroup. (@k82cn)
• #618 Added JobPriority e2e. (@k82cn)
• #622 Update doc & deployment. (@k82cn)
• #626 Fix helm deployment (@thandayuthapani)
• #625 Rename PrioritizeFn to NodeOrderFn (@thandayuthapani)
• #592 Add performance metrics for scheduling (@Jeffwan)

v0.4

Notes:

• Gang-scheduling/Coscheduling by PDB is depreciated.
• Scheduler configuration format and related start up parameter were updated, refer to example for detail.

Issues:

• #534 Removed old design doc to avoid confusion. (@k82cn)
• #536 Migrate from Godep to dep (@Jeffwan)
• #533 The design doc of PodGroup Phase in Status. (@k82cn)
• #540 Examine all pending tasks in a job (@Jeffwan)
• #544 The design doc of "Dynamic Plugin Configuration" (@k82cn)
• #525 Added Phase and Conditions to PodGroup.Status struct (@Zyqsempai)
• #547 Fixed status protobuf id. (@k82cn)
• #549 Added name when register plugin. (@k82cn)
• #550 Added SchedulerConfiguration type. (@k82cn)
• #551 Upgrade k8s dependencies to v1.13 (@Jeffwan)
• #535 Add Unschedulable PodCondition for pods in pending (@Jeffwan)
• #552 Multi tiers for plugins. (@k82cn)
• #548 Add e2e test for different task resource requests (@Jeffwan)
• #556 Added VolumeScheduling. (@k82cn)
• #558 Reduce verbosity level for recurring logs (@mateusz-ciesielski)
• #560 Update PodGroup status. (@k82cn)
• #565 Removed reclaim&preempt by default. (@k82cn)

Release Notes:

• #499 Added Statement for eviction in batch. (@k82cn)
• #491 Update nodeInfo once node was cordoned (@jiaxuanzhou)
• #490 Add version options for kube-batch (@jiaxuanzhou)
• #489 Added Statement. (@k82cn)
• #480 Added Pipelined Pods as valid tasks. (@k82cn)
• #468 Detailed 'unschedulable' events (@adam-marek)
• #470 Fixed event recorder schema to include definitions for non kube-batch entities (@adam-marek)
• #466 Ordered Job by creation time. (@k82cn)
• #465 Enable PDB based gang scheduling with discrete queues. (@adam-marek)
• #464 Order gang scheduled jobs based on time of arrival (@adam-marek)
• #455 Check node spec for unschedulable as well (@jeefy)
• #443 Checked pod group unschedulable event. (@k82cn)
• #442 Checked allowed pod number on node. (@k82cn)
• #440 Updated node info if taints and lables changed. (@k82cn)
• #438 Using jobSpec for queue e2e. (@k82cn)
• #433 Added backfill action for BestEffort pods. (@k82cn)
• #418 Supported Pod Affinity/Anti-Affinity Predicate. (@k82cn)
• #417 fix BestEffort for overused (@chenyangxueHDU)
• #414 Added --schedule-period (@k82cn)

Docker Image:

docker pull kubesigs/kube-batch:v0.3

Features

Features

Docker Images

docker pull kubesigs/kube-batch:v0.2

Features

Docker Images

docker pull kubesigs/kube-batch:v0.2