Intel Device Plugins for Kubernetes v0.31.1

What's Changed

• v0.31.1 fixes a bug where the operator seamless upgrade for pinned images (specified in the OLM bundle) where not upgraded. The bug only impacts users via operatorhub.io/Openshift.

Bugs fixed

• #1873

Full Changelog: v0.31.0...v0.31.1

Intel Device Plugins for Kubernetes v0.31.0

Highlights

The release adds a new image intel-gpu-levelzero, improves operator to capability level 3 and adds CDI support in selected plugins.

Generic

• Update k8s to v1.31.0
• Build release images using the latest Go version (v1.23.x)
• Add CDI support
• Pin workload actions to SHAs
• Sign published containers

Bugs fixed

• #1779
• #1836
• #1840
• Fix QAT heartbeat parsing (#1796)

FPGA

• Move to CDI hooks to enable orchestration programmed mode also with containerd.

Operator

• Improve Operator's capability level to 3

GPU

• Initial support for CDI
• Add a new image that allows retrieving GPU details from LevelZero/Sysman
• Support WSL

Pull requests (excluding dependabot)

• xpumanager sidecar: add note about using HTTPS with xpum by @tkatila in #1738
• qat: improve qat_dpdk_app, openssl-qat-engine by @hj-johannes-lee in #1685
• Post 0.30.0 updates by @tkatila in #1739
• demo: opencl-icd: fix repo key install by @tkatila in #1744
• FPGA: support CDI by @bart0sh in #1745
• pkg/deviceplugin: move to grpc.NewClient() by @mythi in #1748
• Pin workflow actions sha's by @tkatila in #1741
• qat: drop c6xxvf from defaults by @mythi in #1751
• idxd: include accel-config sources in the image by @tkatila in #1761
• demo: accel-config-demo updates by @mythi in #1764
• go.mod k8s.io updates and buildah fix by @mythi in #1783
• fpga: add subsystem vendor & device fields to PCIDevice by @jstamel in #1768
• add updateStrategy to daemonSets by @hj-johannes-lee in #1786
• IAA/DSA test updates by @mythi in #1795
• QAT: make plugin read trimmed heartbeat status by @hj-johannes-lee in #1796
• misc version updates by @mythi in #1807
• TLS cipher updates by @tkatila in #1808
• workflow: sign release containers by @tkatila in #1806
• tls: drop additional ciphers by @tkatila in #1811
• build: use Go 1.23.0 by @mythi in #1814
• xpumanager sidecar: verify certificate with HTTPS by @tkatila in #1816
• Restructure CDI use with plugins by @tkatila in #1810
• GPU-plugin: add cdi support by @tkatila in #1827
• workflow: add malware scan for publish by @tkatila in #1829
• Misc fixes vol. 9/2024 by @tkatila in #1835
• Gpu levelzero sidecar by @tkatila in #1803
• levelzero: add missing calloc return value check by @tkatila in #1844
• IAA/DSA config updates by @mythi in #1849
• Add cpu/memory requests and limits by @hj-johannes-lee in #1819
• gpu: add a note about non-default namespaces with fractional resources by @tkatila in #1850
• operator: kube-rbac-proxy to 0.18.1 by @tkatila in #1853
• workflow: publish: enable gpu-levelzero by @tkatila in #1854

New Contributors

• @jstamel made their first contribution in #1768

Full Changelog: v0.30.0...v0.31.0

Intel Device Plugins for Kubernetes v0.30.0

Highlights

The release adds new functionality to GPU and QAT plugins, and adds toleration support to xDevicePlugin CRDs.

Generic

• Update k8s to v1.30.0
• Build release images using the latest Go version (v1.22.3)
• CI running on a bare metal Intel® Xeon Gen 4 node

Bugs fixed

• #1676
• #1688

Operator

• Add tolerations to xDevicePlugin CRDs

GPU

• Add support for the new xe KMD

QAT

• Add support for 420xx QAT devices
• Add AppArmor unconfided annotation configurability in the operator

Pull requests (excluding dependabot)

• Post 0.29.0 updates by @tkatila in #1635
• e2e: sgx: fetch SGX DC attestation primitives as tar.gz by @tkatila in #1636
• doc+test image updates by @mythi in #1637
• privileged is not needed for openshift by @mregmi in #1640
• e2e: fpga: fix intermittent failures by @tkatila in #1639
• qat: Add AppArmor unconfided anntotation configurability in the operator by @ozhuraki in #1591
• Introduce bug,feature request and support templates by @tkatila in #1647
• FPGA: Drop OpenCL UUID bitstream validation by @eliaskoromilas in #1643
• Fix false failures with trivy config scan by @tkatila in #1652
• chore: Fix multiple typos by @hugo-syn in #1653
• minor version updates by @mythi in #1662
• Go 1.22 + version updates by @mythi in #1660
• openssl-qat-engine: update versions by @mythi in #1672
• demo: update idxd-config to v4.1.6 by @mythi in #1680
• webhooks: make SGX mutator registration to follow other webhooks by @mythi in #1677
• GPU: Add support for the new xe KMD by @tkatila in #1670
• Operator: add support for tolerations and add differentiation labels by @tkatila in #1686
• go.mod: upgrade to k8s.io v1.30 release by @mythi in #1698
• ci: update envtest k8s API versions by @mythi in #1714
• images: update toybox to v0.8.11 by @mythi in #1716
• Use SPR BM for e2e by @tkatila in #1699
• workflow: disable codecov report by @tkatila in #1722
• crypto-perf: update dpdk version to 24.03 by @tkatila in #1721
• Add QAT support for 420xx driver and the 4946 device (GNR-D) by @tkatila in #1719
• ci: move to golangci-lint v1.57.2 by @mythi in #1720
• minor version updates by @mythi in #1727
• Fix repo for opencl-icd, workaround build issues and cherry-pick openssl-qat-engine changes by @tkatila in #1728

New Contributors

• @eliaskoromilas made their first contribution in #1643
• @hugo-syn made their first contribution in #1653

Full Changelog: v0.29.0...v0.30.0

Intel Device Plugins for Kubernetes v0.29.0

Highlights

The release includes various fixes and updates to components.

Generic

• Update k8s to v1.29.0
• Build release images using the latest Go version (v1.21.5)
• Use stricter build flags
• Updates to workflows and new scorecards
• Build images with buildkit
• Move towards using ktesting/textlogger

Bugs fixed

• #1418
• #1556
• #1604

NFD

• Update SGX deployment to use NFD labels
• Add vfio-pci dependency to QAT rules

Operator

• Allow multiple objects per CR
• Fix operator updating daemonset when CR has not changed

GPU

• Change Level Zero's default affinity mask format when tiles are used, and do not set mask with 1-tile devices
• Set monitoring enabled in the default deployment
• Documentation updates

QAT

• Support QAT heartbeat in >=6.6 kernels
• Add new QAT resources (capabilities)

Fake GPU Device

• Support xelink labeling

VPU

• VPU source code removed from repository

Pull requests (excluding dependabot)

• docs + operator updates to 0.28.0 by @tkatila in #1547
• build: move to Go 1.21.x by @mythi in #1508
• vpu: remove deprecated plugin by @mythi in #1549
• qat: heartbeat support + new capabilities by @tkatila in #1559
• Cleanups by @mythi in #1572
• Misc fixes by @tkatila in #1561
• Update versions (dlb demo and cert-manager, kind) by @hj-johannes-lee in #1543
• controller cleanups by @mythi in #1583
• ci: update golangci-lint by @mythi in #1586
• build: use stricter build flags by @tkatila in #1593
• operator: remove one-cr-per-kind limitation by @tkatila in #1579
• workflows: add scorecard and codeql by @tkatila in #1597
• sgx: use NodeFeatureRule SGX label in the NFD overlay by @mythi in #1596
• xelink support by @hsyrja in #1580
• QAT test updates by @mythi in #1599
• logging: move away from klogr to ktesting/textlogger by @mythi in #1600
• go.mod: upgrade to k8s.io v1.29 by @mythi in #1584
• Minor improvements to "render-device.sh" GPU plugin helper script by @eero-t in #1605
• codeql fixes by @mythi in #1606
• workflow: run codeql for PRs and ignore actions for doc changes by @tkatila in #1608
• Fix last of codeql issues by @tkatila in #1610
• images: use buildkit again by @mythi in #1615
• Change GPU plugin's behavior as Level zero's default hierarchy mode changed from composite to flat by @tkatila in #1601
• go.mod: upgrade to k8s.io v1.29 release by @mythi in #1622
• demo: update idxd-config to v4.1.4 by @mythi in #1618
• ci: cleanup linter settings by @mythi in #1626
• nfd rules: require vfio-pci module to be present for qat by @tkatila in #1627

New Contributors

• @hsyrja made their first contribution in #1580

Full Changelog: v0.28.0...v0.29.0

Intel Device Plugins for Kubernetes v0.28.0

Highlights

The release includes various fixes and updates to components.

Generic

• Update k8s to v1.28.2
• Build release images using the latest Go version (v1.20.8)
• NFD updated to 0.14.1

Bugs fixed

• #1471
• #1427
• #1357
• #1506
• Fix crash with GPU plugin related to kubelet connection timeouting

Operator

• Added better support for operator bundle with forced plugin digests
  • Plugin images can be hard coded via environment variables
  • https://github.com/intel/intel-device-plugins-for-kubernetes/tree/main/cmd/operator#upgrade
• Updated to CRD descriptions

GPU

• Functionality from initcontainer moved to GPU plugin and to NFD Feature rules
• Added example tensorflow workload
• Updated documentation

QAT

• Correct a log print verbosity from warning to info
• Updated documentation

E2E

• Restructured code and fixed flows

VPU

• Images will not be published
• VPU plugin and demo code will be removed after 0.28.0 release

Pull requests (excluding dependabot)

• dsa: replace token with read_buffer in conf by @hj-johannes-lee in #1465
• Disable VPU plugin builds by @tkatila in #1456
• demo: crypto-perf, accel-config-demo and openssl-qat-engine updates by @mythi in #1478
• Add SELinux Labels for DSA and IAA by @mregmi in #1479
• Change the regular expression of the device type to support sIOV device by @lzha111 in #1474
• Link fixes, log fix and yaml updates by @tkatila in #1487
• Update references to OpenShift by @hershpa in #1509
• update accel-config version to 4.1 by @hj-johannes-lee in #1516
• tool updates by @mythi in #1524
• images: base image specific toybox config and version updates by @mythi in #1525
• Makefile: update versions & FPGA: fix naked return error from linter by @hj-johannes-lee in #1477
• version updates by @mythi in #1496
• k8s.io 1.28.1 update and SGX rework by @mythi in #1513
• k8s v1.28.2 and NFD updates by @mythi in #1539
• go.mod: update to k8s.io v1.28.0 by @mythi in #1486
• go.mod: bump sigs.k8s.io/controller-runtime from 0.15.1 to 0.16.0 by @mythi in #1507
• e2e: remove duplicate code of testing with operator by @hj-johannes-lee in #1472
• e2e,sgx: divide single It() to have multiple layers by @hj-johannes-lee in #1469
• e2e: fix webhook TLS tests by @mythi in #1488
• e2e: make demo pod logged only when failed running by @hj-johannes-lee in #1470
• e2e,gpu,fpga: divide single It() to have multiple layers by @hj-johannes-lee in #1491
• e2e,dsa,iaa: remove unnecessary layer ginkgo.Describe by @hj-johannes-lee in #1489
• e2e: make running subsets of e2e tests more organized by @hj-johannes-lee in #1512
• e2e: gpu: add a basic tensorflow test by @tkatila in #1374
• gpu: copy nfdhook functionality to gpu-plugin by @tkatila in #1492
• gpu docs update and xpum-sidecar deployment version bump by @tkatila in #1533
• gpu_fakedev: better dra support by @uniemimu in #1518
• Revert "disable gpu e2e test temporarily" by @tkatila in #1501
• operator: add missing IaaDevicePlugin finalizers RBAC by @mythi in #1475
• operator: add image upgrade with env vars by @hj-johannes-lee in #1499
• sgx & gpu crds: improve descriptions by @tkatila in #1534
• sgx: use sgx-sdk-demo app instead of busybox by @hj-johannes-lee in #1464
• webhooks: stop handling Pod updates by @mythi in #1484
• workflows: rework, trivy & publish by @tkatila in #1511
• Workflows/publish fix by @tkatila in #1515
• CI cleanups by @mythi in #1517
• ci: drop kind tests from Jenkins too by @mythi in #1503
• ci: skip linter checks for cmd/vpu_plugin by @mythi in #1473
• envtest for older k8s versions by @mythi in #1541
• qat: update README by @mythi in #1535
• codeowners: add eero-t as reviewer for gpu_fakedev by @tkatila in #1538
• doc: add a release procedures section by @tkatila in #1504

New Contributors

• @lzha111 made their first contribution in #1474
• @hershpa made their first contribution in #1509

Full Changelog: v0.27.1...v0.28.0

Intel Device Plugins for Kubernetes v0.26.2

Highlights

The release improves QAT support and updates to newer k8s versions.

Generic

• Update k8s versions to 1.26.7

Operator

• Fix QAT provisioning config volumeMount #1431
• Increase operator's memory limits to 100/120Mi #1429

QAT

• Add support for 401xx/402xx devices #1434

Full Changelog: v0.26.1...v0.26.2

Intel Device Plugins for Kubernetes v0.27.1

Highlights

The release includes updates to core k8s components, and improves XPU-manager sidecar usability on single tile devices.

Generic

• Update k8s to v1.27.3
• Build release images using the latest Go version (v1.20.5)

Images

• No new images.

Operator

• No updates.

GPU

• XPU-Manager sidecar supports XeLinks with single tile devices and HTTPS.

SGX

• No updates.

DSA

• No updates.

IAA

• No updates.

FPGA

• No updates.

QAT

• No updates.

VPU

• No updates.

DLB

• No updates.

Intel Device Plugins for Kubernetes v0.27.0

Highlights

The release includes various generic fixes and component upgrades. This will be the last release to include VPU-plugin.

Generic

• Lots of dependency updates
  • k8s to 1.27.2
  • controller runtime to 0.15.0.
• Documentation improvements including Helm install instructions
• Use more generic name for NFD features host directory volume
• Remove use of "kube-system" namespace

Images

• No new images.

Operator

• Fix QAT provisioning config volumeMount
• Move to controller-tools 0.12.0
• Increase memory resource limits to 100/120Mi

GPU

• Add error checking for Kubelet flow

SGX

• Remove initcontainer in favor of NFD's node feature rules

DSA

• Initcontainer: Move to Debian's accel-config
• Accel-config-demo: update to accel-config 4.0

IAA

• No updates.

FPGA

• No updates.

QAT

• Add support for 401xx and 402xx VF PCI IDs
• Demo updated to 23.02.0 QAT lib, 1.1.0 QAT engine and 0.5.0 async nginx
• Documentation updates

VPU

• Last release for the VPU-plugin!
• No updates.

DLB

• Update to v8.3.0
• Modify init script to not fail on non-existent sriov_numvfs file

Intel Device Plugins for Kubernetes v0.26.1

Highlights

The release fixes a bug with Environment variables and decreases GPU plugin's demand for the API server.

Generic

• pkg/deviceplugin: do not reset Envs/Annotations from previous loops #1343

Images

• No new images.

Operator

• Update kube-rbac-proxy image to v0.14.1 #1397
• Accept image SHA digests #1390

GPU

• Mount "by-path" directory #1371
• Increase and randomize cleanup interval & leverage kubelet API #1363

SGX

• No updates.

DSA

• No updates.

IAA

• No updates.

FPGA

• No updates.

QAT

• No updates.

VPU

• No updates.

DLB

• No updates.

Intel Device Plugins for Kubernetes v0.26.0

Release notes

Highlights

The release adds support for configuring crypto and compression capabilities through sysfs on QAT Gen4 platforms.

Generic

• Upgraded Kubernetes API to v1.26 level (#1310).

Images

• New image XPU Manager Sidecar (#1280).

Operator

• Added Liveness and Readiness probes (#1232).

GPU

• nfdhook: Fix bug with multi-labels starting or ending with invalid characters (#1285).
• deployment: New deployment variant with monitoring and shared dev options enabled (#1292).

SGX

• No updates.

DSA

• No updates.

IAA

• No updates.

FPGA

• No updates.

QAT

• Added support for configuring crypto and compression capabilities through sysfs on QAT Gen4 platforms (#1234).

VPU

• No updates.

DLB

• Fixed DPDK test application parameters (#1307).