Download Latest Version v1.15.1 source code.zip (5.4 MB)
Home / v1.15.0
Name Modified Size InfoDownloads / Week
Parent folder
README.md 2026-10-10 44.2 kB
1 1 weekly downloads
v1.15.0 source code.tar.gz 2026-10-10 4.7 MB
1 1 weekly downloads
v1.15.0 source code.zip 2026-10-10 5.4 MB
1 1 weekly downloads
Totals: 3 Items   10.1 MB 3

Features

  • add aws specific feature gates implementation with --aws-feature-gates flag (#9482) #9482 (Joshua Guo)
  • CEL expression support for maxPodsExpression, kubeReserved, and systemReserved (#9326) #9326 (Zara)
  • Extending KubeletConfiguration struct fields (#9461) #9461 (Zara)
  • expose security group refresh interval as configurable setting (#9526) #9526 (Derek Frank)
  • metrics: bump to the label-aware core and document provider metric dimensions (#9595) #9595 (Derek Frank)
  • add EC2NodeClass support for Nitro Enclaves #8165 (Sebestien Palmerio)
  • decouple reserved offering Available from ReservationCapacity (#9625) #9625 (Derek Frank)
  • add Helm values and docs for the PodDeletionCostManagement feature gate (#9653) #9653 (Derek Frank)
  • implement CloudProvider.Reboot via ec2:RebootInstances (#9676) #9676 (Sarthak Umarani)
  • reason-aware default repair policies with bounded termination grace periods (#9680) #9680 (Sebestien)
  • nvidia and efa driver dra support (#9550) #9550 (Joshua Guo)
  • bump karpenter to support --legacy-node-repair, add settings.legacyNodeRepair to the chart, and document it (#9709) #9709 (Derek Frank)
  • add the RebootForRepair AWS feature gate for reboot repair policies (#9716) #9716 (Sarthak Umarani)

Bug Fixes

Documentation

  • update broken link to Node Repair heading on Disruption Concepts page (#8993) #8993 (Seth McCombs)
  • add g6f fractional GPU NodeOverlay workaround example (#9395) #9395 (Zaid Ahmed Farooq)
  • add conceptual documentation for NodeClaims (#9125) #9125 (Sergiy)
  • add version support policy (#9313) #9313 (Sergiy)
  • RFC for CEL expressions in kubelet config fields (#9342) #9342 (Zara)
  • fix invalid eksctl flag in Zonal Shift onboarding (#9495) #9495 (Sandipan Panda)
  • consolidate status condition, NodeOverlay, and minValues fixes (#9500) #9500 (Ryan Mistretta)
  • add status condition, NodeOverlay, and minValues fixes to /docs (#9502) #9502 (Ryan Mistretta)
  • CEL Expressions and Extended Kubelet Configuration feature site docs (#9508) #9508 (Zara)
  • update for patch release v1.14.1 (#9533) #9533 (Ryan Mistretta)
  • add task for monitoring API throttling (#9534) #9534 (Derek Frank)
  • add ca rotation warning (#9522) #9522 (Joshua Guo)
  • note the reason metric label casing change in the 1.15 upgrade guide (#9581) #9581 (Derek Frank)
  • metrics: docs generator reads []opmetrics.Label and Stage; regenerate metrics.md (#9604) #9604 (Derek Frank)
  • note the 1.15 breaking cloudprovider error label change ("" -> "unknown") (#9600) #9600 (Derek Frank)
  • recommend MostAllocated (#9603) #9603 (Derek Frank)
  • document node repair as a voluntary disruption method (#9635) #9635 (Derek Frank)
  • add kubectl examples to inspect NodePools, EC2NodeClasses, and NodeClaims (#9588) #9588 (Daniel Bae)
  • document terminate-first disruption for drift and repair (#9655) #9655 (Derek Frank)
  • clarify DaemonSet overhead is used for sizing, not reserved (#9442) #9442 (Vincent Creusot)
  • fix typos in concepts documentation (#9429) #9429 (Archong Liu)
  • Node Pool -> NodePool (#9651) #9651 (Ben Bodenmiller)
  • note that metric dimensions can be empty depending on object state (#9657) #9657 (Derek Frank)
  • add guidance for exposing instance configuration through offerings (#9640) #9640 (Jason Deal)
  • generate a well known annotations reference from wellknown.Annotations (#9669) #9669 (Derek Frank)
  • clarify NodeOverlay hourly price units (#9478) #9478 (Mia Miao)
  • instance-family plus instance-size can collapse to one SKU (#9556) #9556 (Andrey Maltsev)
  • restricted runtime labels do not trigger provisioning (#9555) #9555 (Andrey Maltsev)
  • warn about consolidation candidate ordering change in 1.14 upgrade guide (#9695) #9695 (Derek Frank)
  • regenerate the metrics reference for the reboot disruption decision #9691 (Sarthak Umarani)
  • add 1.15 consolidation_type upgrade note, security group refresh interval, and consolidateAfter default (#9715) #9715 (Derek Frank)
  • order the 1.15.0 upgrade guide by impact and document the ordering in AGENTS.md (#9712) #9712 (Derek Frank)
  • document the do-not-repair DisruptionBlocked event (#9698) #9698 (GaneshBannur)

Performance Improvements

Tests

  • bump IODCR interruption notice period (#9543) #9543 (Ryan Mistretta)
  • drop stale testing for unmanaged and managed instance profile (#9545) #9545 (Ryan Mistretta)
  • Adding E2E Tests for Zonal Shift provider. (#9221) #9221 (sagdhana)
  • use g.Expect in ExpectInterruptibleAndSourceCapacityCanceled (#9609) #9609 (Ryan Mistretta)
  • bump upgrade suite from_git_ref pin (#9607) #9607 (Drew Sirenko)
  • fix IPv6 e2e by disabling vpc-cni subnet discovery (#9608) #9608 (Drew Sirenko)
  • compare nodes by UID in e2e Monitor created/deleted node tracking (#9631) #9631 (Derek Frank)
  • run core disruption and repair suites upstream and fix repair e2es for voluntary repair (#9647) #9647 (Derek Frank)
  • cover terminate-first drift and repair on a full capacity reservation #9654 (Sebestien Palmerio)
  • use the initialized node in repair condition e2es (#9675) #9675 (Sebestien)
  • fit repair e2es inside their timeouts now that core clamps condition transition times to node creation (#9687) #9687 (Derek Frank)
  • inject reason-matched repair faults so repair e2es hit the 10 minute XID policy #9692 (Derek Frank)
  • run the reboot e2e specs on EC2 #9691 (Sarthak Umarani)
  • move repair and reboot e2es into their own Repair suite and bump Go to 1.26.9 for vulncheck (#9699) #9699 (Derek Frank)
  • use >1 vCPU instances and a 20m timeout for Windows kubelet e2e specs (#9701) #9701 (Ryan Mistretta)
  • e2es: enable DRA NVIDIA/EFA feature gates and stop ignoring DRA requests (#9708) #9708 (Ryan Mistretta)
  • fall back across random instance types and zones when creating E2E capacity reservations (#9707) #9707 (Derek Frank)
  • fix flaky zonal shift test #9710 (Ryan Mistretta)
  • add back in legacy repair e2es (#9714) #9714 (Ryan Mistretta)
  • e2e for terminate-first static repair and drift with a full capacity reservation (#9706) #9706 (Derek Frank)

Continuous Integration

  • unit test and ci fix for dl1.24xlarge does not exist error (#9343) #9343 (Joshua Guo)
  • pin all release/CI build tooling to explicit versions (#9396) #9396 (Derek Frank)
  • fix flaky should use same launch template (#9459) #9459 (Ryan Mistretta)
  • fix the website preview workflow (#9515) #9515 (Joshua Guo)
  • fixing the broken fixed website preview workflow (#9530) #9530 (Joshua Guo)
  • harden website preview workflow's artifact handling (#9587) #9587 (GaneshBannur)
  • collapse per-suite periodic e2e Slack pings into one run summary (#9589) #9589 (Drew Sirenko)
  • Add read perms to e2e-matrix-trigger (#9596) #9596 (Drew Sirenko)
  • install the go.mod Go version in the private cluster CodeBuild job (#9671) #9671 (Derek Frank)
  • bump eksctl to v0.217.0 for E2E (#9672) #9672 (Derek Frank)
  • upgrade from the latest release in the E2E Upgrade suite (#9677) #9677 (Derek Frank)
  • keep the GitHub "Latest" release badge on the newest version (#9397) #9397 (Derek Frank)
  • fix private cluster e2e helm, prometheus webhook image, and zonal shift endpoint (#9679) #9679 (Derek Frank)
  • run required checks in the merge queue (#9681) #9681 (Derek Frank)
  • deploy website previews on /karpenter preview instead of every PR push (#9686) #9686 (Derek Frank)
  • give the IPv6 vpc-cni addon a pod identity so eksctl v0.217.0 can create E2E clusters (#9688) #9688 (Derek Frank)
  • use IRSA for Karpenter on IPv6 E2E clusters so eksctl skips the vpc-cni pod identity (#9689) #9689 (Derek Frank)

Chores

Commits

  • 90f724a: split scheduling test suite into two (Sagar Dhana) #9341
  • Update label naming for consistency (#9410) #9410 (ahreehong)
  • 4a562f4: Add memory-manager-reserved-memory to BottlerocketKubernetes settings (#9427) (Tanvir Tatla) #9427
Source: README.md, updated 2026-10-10