Sign Up
Log In
Log In
or
Sign Up
Places
All Projects
Status Monitor
Collapse sidebar
home:MaxxedSUSE
nvidia-container-toolkit
nvidia-container-toolkit.changes
Overview
Repositories
Revisions
Requests
Users
Attributes
Meta
File nvidia-container-toolkit.changes of Package nvidia-container-toolkit
------------------------------------------------------------------- Mon Oct 10 06:21:54 UTC 2022 - Michal Jura <jura@suse.com> - Update to version 1.11.0 (jsc#SLE-18750): * Ensure that base package is built for debian * Update libnvidia-container submodule * Bump version to v1.11.0 * Update git commit command * Add release tests for fedora35 * Clean up repo test scripts * Add fedora35 to release and signing scripts * Ensure CLI versions are set correctly for RPM packages * Add changelog for 1.11.0-rc.3 * Update libnvidia-container * Update CUDA base image to 11.7.1 * Update libnvidia-container submodule * Increase package build timeout to 3 hours for slow aarch64 builds * Use single config file for centos, al2, and fedora * Add fedora35 CI targets * Add fedora targets to release scripts * Add fedora35 package targets * Switch to single docker file yum-based rpm builds * Use new packages in toolkit image * Split nvidia-container-toolkit package * Fix centos8 test image * Fix indentation in makefile * Update vendoring * Specify hook structure instead of importing Podman * Fix cleanup of nvidia-container-toolkit link * Use proper cuda image for containerd tests * Update subcomponents * Update image used for containerd tests * Output applied config to toolkit container stdout * Ensure that toolkit-container sets correct default value * Fix setting of toolkit config option in toolkit container * Update libnvidia-container * Update vendoring * Use nvinfo package from go-nvlib * Add modifier to inject Tegra platform files * Bump version to 1.11.0-rc.3 * Fix setting of LIBNVIDIA_CONTAINER_TAG * Add CHANGELOG entry for 1.11.0-rc.2 * Update libnvidia-container * Allow accept-visible-devices config options to be set * Remove unused TOOLKIT_ARGS / --toolkit-args * Set toolkit root as flag * Rename toolkitDir toolkitRoot * Move global toolkitDir to options struct * Move toolkit options to struct * Bump version to 1.11.0-rc.2 * Add changelog entries for 1.11.0-rc.1 * Apply 1 suggestion(s) to 1 file(s) * Add root to mounts type * Make error message clearer * Remove Relative method from Locator * Fix bug where ldcache may not contain symlinks * Add tests for identifying libraries * Add nvidia-ctk runtime configure command * Move docker config handling to internal package * Ensure that CDI registry is refreshed * Add runtime config option for CDI spec dirs * Reuse check for existing hook * Update package descriptions and URLs * Update package definitions * Update references to nvidia-container-runtime-hook * Rename -toolkit executable to -runtime-hook * Skip packages that already exist * Use centos:stream8 image for signing * Use device host path to determine properties * Update vendored runc version * Update cdi package and run go mod vendor * Add support for specifying devices in annotations * Add cdi mode to NVIDIA Container Runtime * The licenses make target should not be a check target * Add charDevices discoverer for devices * Create single discoverer per mount type for CSV * Add tooling to check go licenses * Rename discover.NewList to discover.Merge * Add Relative function to Locator interface * Use CUDA.DevicesFromEnvvar to check if modifications are required * Add DevicesFromEnvvars function to CUDA image * Add /etc/cufile.json to list of required mounts * Create GDS and MOFED modifiers * Add discovery of GDS and MOFED devices * Allow globs in filenames for locators * Move cmd/nvidia-container-runtime/modifier package to internal/modifier * Use modifier list and discoverModifer * Add lists of modifiers to allow for modifier compositioning * Ensure test/output path exists * Update vendoring * Update nvidia-docker and nvidia-container-runtime * Update nvidia-docker and nvidia-container-runtime branches to main * Allow any 1.* version of libnvidia-container package * Switch to latest docker and docker dind in CI * Allow libnvidia-container1 version to be specified directly * Update build scripts to set libnvidia-container version * Bump version to 1.11.0-rc.1 * Update libnvidia-container submodule to v1.10.0 * Bump version to v1.10.0 * Update toolkit images to use NGC DL license * Bump nvidia-docker version * Switch default container-toolkit image target to ubuntu20.04 * Only generate amd64 images for ubuntu18.04 * Remove build and release of centos8 container-toolkit images * Use ubi8 base image for centos8 * Bump CUDA base image version to 11.7.0 * Update config files with options and defaults * Update NVIDIA Container Runtime readme * Update libnvidia-container * Bump version to 1.10.0-rc.4 * Also set default_runtime.options.BinaryName * Also cleanup v1 default_runtime if BinaryName is set * Also set Runtime file v1 containerd runtime config * Use BinaryName for v1 containerd runtime config * Update libnvidia-container * Return default config if config path is not found * Ignore NVIDIA_REQUIRE_JETPACK* for image requirements * Fix bug in tegra detection * Fix assertCharDevice matching on all files * Include git commit in changelog URL * Automatically generate changelogs in docker builds * Add dummy entry for rpm changelog matching other components * Format CHANGELOG.md as markdown * Move debian changelog to CHANGELOG.md * Update libnvidia-container version * Bump version to 1.10.0-rc.3 * Update libnvidia-container * Update changelog for release * Ensure that git commit is set in docker build * Set the version and gitCommit in the Makefile * Add version output to CLIs * Call logger.Reset() to ensure errors are captured * Skip setting of log file for --version flag * Include HasNVML check in ResolveAutoMode * Add HasNVML function to check if NVML is supported * Remove unneeded legacy discovery * Remove --force flag from nvidia-container-runtime-hook * Replace experimental and discover-mode * Move ResolveAutoMode to info package * Move isTegraSystem to internal info package * Update nvidia-container-runtime config options * Use toml unmarshal to read runtime config * Add hook to create specific links * Add --link option to nvidia-ctk hook create-symlinks command * Factor linkCreation into method * Improve symlink creation loop * Use singular instead of plural for hook arguments * Use executable locator to find low-level runtime * Use lookup.GetPath from runtime hook * Add lookup.GetPath and lookup.GetPaths functions * Use state.GetContainerRoot in nvidia-ctk hook subcommands * Add GetContainerRoot to oci.State type * Support runc logging command line options * Make output of bundle directory a debug message * Switch to debug logging when locating runtimes * Add nvidia-container-runtime.runtimes config option * Fix form -> from in comment * Add debug logging when checking requirements * Add compute capability of first device as arch property * Add CUDA ComputeCapability function * Add debug log for command line arguments * Return low-level runtime if subcommand is not create * Check requirements before creating CSV discoverer * Add processing for requirements and constraints * Return raw spec from Spec.Load * Add basic CUDA wrapper * Use CUDA image abstraction for runtime hook * Add CUDA image abstraction * Add gcc for Amazonlinux builds * Use go install to install go development tools * Bump golang version to 1.17.8 * Update go vendoring * Fix image building due to GPG key update * Use semver package to parse CUDA version * Update libnvidia-container reference * libnvidia-container: 'main' track branch * Remove dockerhub publishing * Bump github.com/containers/podman/v4 from 4.0.1 to 4.0.3 * Improve handling of git remotes for gh-pages packages * Add scripting to sign and publish packages * Add envvar for package versions * Rename release.sh to build-packages.sh * Change master references to main * Update libnvidia-container submodule * Bump version to v1.10.0-rc.2 * Add commented experimental option to config files * Revert "[ci] Skip external releases if associated OUT_REGISTRY value is empty." * Revert "[ci] echo skipped commands" * Update libnvidia-container * Add log-level config option for nvidia-container-runtime * Remove exsiting NVIDIA Container Runtime Hooks from the spec * Specify --force flag when invoking nvidia-container-runtime-hook * Raise error if hook invoked in experimental mode without force flag * Export GetDefaultRuntimeConfig * Make order of discoverers deterministic * Refactor CSV discovery to make char device discovery clearer * Fix creation of CSV parser in create-symlinks * Fix creation of CSV parser in create-symlinks * Move NVIDIA Container Runtime Hook executable name to shared constant * Use DefaultExecutableDir to determine default paths * Refactor CSV file parsing * Add missing close when reading CSV file * Return unmodified runtime if specModifier is nil * Inject symlinks hook for creating symlinks in a container * Add create-symlinks subcommand to create symlinks in container for specified CSV files * Move reading of container state for internal/oci package * FIX: Rename containerSpec flag to container-spec * Include nvidia-ctk in deb and rpm packages * Add cache for mounts * Add discovery for ldconfig hook that updates the LDCache * Add nvidia-ctk config section * Add hook command to nvidia-ctk with update-ldcache subcommand * Add stub nvidia-ctk CLI * Refactor hook creation * Add auto discover mode and use this as the default * FIX: Rename DefaultRoot to DefaultMountSpecPath * FIX: Improve locator map construction * FIX: Update TODO for container path * FIX: Use MountSpec* constants * FIX: Remove unused NewFromCSV constructor * Correct typo in constructor name * Add support for NVIDIA_REQUIRE_JETPACK envvar * Add csv discovery mode to experimental runtime * Add CSV-based discovery of device nodes * Add CSV-based discovery of mounts * Add locators for symlinks and character devices * Add code to process Jetpack CSV files * FIX: Make isNVIDIAContainerRuntimeHook mode idiomatic * FIX: Simplify hook remover * FIX: Rename path locator as executable locator * FIX: Rename CLIConfig to ContainerCLIConfig * FIX: Factor out specModifier construction into function * FIX: Don't log that hooks is being removed if it is not * FIX: Fix typo in comment * [ci] echo skipped commands * Fix typo in variable name * Add basic README for nvidia-container-runtime * Make error logging less verbose by default * Implement hook remover for existing nvidia-container-runtime-hooks * Read top-level config to propagate Root to experimental runtime * Split loading config from reader and getting config from toml.Tree * Implement experimental modifier for NVIDIA Container Runtime * Add stable discoverer for nvidia-container-runtime hook * Add lookup abstraction for locating executable files * Move runtime config to internal package * Don't skip internal packages for linting * Add experimental option to NVIDIA Container Runtime config * Update libnvidia-container * [ci] Skip external releases if associated OUT_REGISTRY value is empty. * Move modifier code for inserting nvidia-container-runtime-hook to separate package * Import modifying runtime abstraction from experimental runtime * Add test package with GetModuleRoot and PrependToPath function * Ensure that Exec error is also logged to file * Update go vendoring * Update podman hooks dependency * Add .shell make target for non-Linux development * Add gcc for centos package builds including cgo * Update gitignore * Switch from centos:8 to centos:stream8 images to build centos8 packages * Update git submodules * Update libnvidia-container submodule to v1.10.0-rc.1 * Bump version to 1.10.0-rc.1 * Use nvcr.io registry for Ubuntu CUDA base images * Add CI definitions for building and publishing Ubuntu20.04 images * Upcate libnvidia-container submodule * Bump version to 1.9.0 * Update libsasl in both ubuntu/ubi toolkit images to address CVE-2022-24407 * Update libnvidia-container subcomponent * Use 'none' instead of 'NONE' to skip containerd restart * Add --restart-mode to docker config CLI * Update component submodules * Fix pushing of short tag for devel images * Add multi-arch image scans * Also search /usr/lib/aarch64-linux-gnu for libnvidia-container libs * Enable multi-arch builds in CI * Enable multi-arch builds * Allow buildx to be used for mulit-arch images * Rename TARGETS make variable to DISTRIBUTIONS * Specify docker platform args for build and run commands * Ensure that Ubuntu20.04 images also build * Remove unneeded build-all CI steps * Fix centos8 builds * Update submodules * Remove unneeded build-all CI steps * Update submodules * Fix centos8 builds * Bump version to 1.9.0-rc.1 * Update centos:8 mirrors for release tests * Update libnvidia-container submodule * Update changelogs * Update libnvidia-container submodule * Bump version to 1.8.1 * Fix changelog entry in rpm spec * Update component submodules * Bump version to 1.8.0 * Use 2h30m timeout for all packaging stages * Update centos8 mirrors * Update sub-modules * Bump version to 1.8.0-rc.3 * Update libnvidia-container submodule * Update components before building release * Copy libnivida-container-go.so to toolkit directory * Remove support for amazonlinux1 * Add check for matching toolkit and lib versions to release script * Update git submodules * Bump version to 1.8.0-rc.2 * Update CUDA image version to 11.6.0 * Update libnvidia-container submodule for WITH_NVCGO CI build fix * Update libnvidia-container submodule * Update nss on centos7 to address CVEs * Allow packages to be specified to address CVEs * Update libnvidia-container submodule * Bump version post 1.7.0 release * Enable release of toolkit-container images * Simplify skipping of scans * Add delay and timeout to image pull job * Pull public staging images to scan and release * Address review comments * Add script to pull packages from packaging image * Add placeholder for testing packaging image * Add packaging target to CI * Add packaging target that includes all release packages * Include all architecture packages in toolkit container * Update libnvidia-container submodule * Bump version to 1.7.0 * Update submodules * Bump golang version to 1.16.4 * Add versions.mk file to define versions * Specify nvidia-container-runtime and nvidia-docker versions * Bump post 1.7.0-rc.1 release * Don't rebuild packages for every local run * Add basic multi-arch support to release tests * Rework init repo for centos8 release tests * Update libnvidia-container dependency for release * Update changelog * Update vendoring * Specify containerd runtime type as string * Update submodules * Override LIB_TAGS for runtime and docker wrapper * Bump post 1.6.0 release * Add jetpack-specific config.toml * Specify config.toml file suffix as docker build arg * Add nvidia-container-config option to overide drivercapabilities * Update components versions for 1.6.0 release * Bump version to 1.6.0 * Fix logging to stderr instead of file logger * [ci] remove --pss flag from pulse scanning * Check for matching tags in release script * Get tags for all components in get-component-versions script * Update submodules * Update libnvidia-container to ff6ed3d5637f0537c4951a2757512108cc0ae147 * Update libnvidia-container submodule to 1.6.0-rc.3 * Bump version post v1.6.0-rc.2 release * Update submodules for packaging * Update nvidia-docker submodule * Auto update debian changelog and release date * Update nvidia-docker submodule * Forward nvidia-container-toolkit versions to dependants * Update libnvidia-container submodule * Update nvidia-container-runtime submodule * Update imported OCI runtime spec * Add basic test for preservation of OCI spec under modification * [ci] use pulse instead of contamer for scans * Import cmd/nvidia-container-runtime from experimental branch * Remove unneeded files * Import internal/oci package from experimental branch * Rename RELEASE_DEVEL_TAG for consistency * Remove rule for merge requests * Add changelog entry for config.json path changes * Bump version to 1.6.0-rc.2 * skip error when bundleDir not exist * Remove rule for merge requests * Add internal CI definition for release * Add CI to build toolkit-container image * Add dockerfile and makefile to build toolkit-container * Copy container test scripts from container-config * Copy cmd from container-config * Update go vendoring * Update submodules * Add aarch64 build for Amazon Linux 2 * Extend release testing toolking to allow for upgrade testing * Use build image directly in CI * Update DEVELOPMENT.md * Update submodules for packaging fixes * Add docker-based tests for package installation workflows * FIXUP: Update development * Require at least a matching libnvidia-container-tools version * Use consistent package revisions for all rpm-based packages * Fix nvidia-container-runtime breaks / replaces dependency * Add release script for specific targets * Remove docker-all target from Makefile * Add basic version checks * Add nvidia-docker as a git submodule * Add nvidia-container-runtime as a git submodule * Add libnvidia-container as a git submodule * Apply edits for the NVIDIA container toolkit * Copy README.md from nvidia-docker * Copy scripts from nvidia-container-toolkit-release * Bump version for 1.6.0 development * Update debian and rpm package definitions * Add PREFIX make variable to control command output * Make all commands and copy executables * Add cmds target to makefile to build all go commands * Update go vendoring * Update package references * Copy code from nvidia-container-runtime * Bump version for post 1.5.1 development * Revert "Add support for NVIDIA_FABRIC_DEVICES" * Revert "Bump version to 1.6.0~rc.1" * Bump version to 1.6.0~rc.1 * Add support for NVIDIA_FABRIC_DEVICES * Use extends keyword for build-one and build-all * Improve CI for container toolkit ------------------------------------------------------------------- Wed Jun 30 12:08:25 UTC 2021 - Michal Jura <mjura@suse.com> - Update to version 1.5.1: * Bump to version 1.5.1 * Explicitly set GOOS when building binary * Add coverage step to CI * Move pkg to cmd/nvidia-container-toolkit * Run go mod vendor and go mod tidy * Fix bug where docker swarm device selection is overriden by NVIDIA_VISIBLE_DEVICES * Use require package for tests * Add coverage to go tests * Update vendoring * Bump version to 1.5.0 ------------------------------------------------------------------- Wed Apr 8 14:34:41 UTC 2020 - Sascha Grunert <sgrunert@suse.com> - Add initial version
Locations
Projects
Search
Status Monitor
Help
OpenBuildService.org
Documentation
API Documentation
Code of Conduct
Contact
Support
@OBShq
Terms
openSUSE Build Service is sponsored by
The Open Build Service is an
openSUSE project
.
Sign Up
Log In
Places
Places
All Projects
Status Monitor
