Skip to content

Quickstart and Usage

Install

Recommended installation method is with Brew:

Terminal window
brew tap defenseunicorns/tap && brew install uds

UDS CLI Binaries are also included with each Github Release

Contributing

Build instructions and contributing docs are located in CONTRIBUTING.md.

Quickstart

The UDS-CLI’s flagship feature is deploying multiple, independent Zarf packages. To create a UDSBundle of Zarf packages, create a uds-bundle.yaml file like so:

kind: UDSBundle
metadata:
name: example
description: an example UDS bundle
version: 0.0.1
packages:
- name: init
repository: ghcr.io/defenseunicorns/packages/init
ref: v0.33.0
optionalComponents:
- git-server
- name: podinfo
repository: ghcr.io/defenseunicorns/uds-cli/podinfo
ref: 0.0.1

The above UDSBundle deploys the Zarf init package and podinfo.

The packages referenced in packages can exist either locally or in an OCI registry. See here for an example that deploys both local and remote Zarf packages. More UDSBundle examples can be found in the src/test/bundles folder.

Declarative Syntax

The syntax of a uds-bundle.yaml is entirely declarative. As a result, the UDS CLI will not prompt users to deploy optional components in a Zarf package. If you want to deploy an optional Zarf component, it must be specified in the optionalComponents key of a particular package.

First-class UDS Support

When running deploy,inspect,remove, and pull commands, UDS CLI contains shorthand for interacting with the Defense Unicorns org on GHCR. Specifically, unless otherwise specified, paths will automatically be expanded to the Defense Unicorns org on GHCR. For example:

  • uds deploy unicorn-bundle:v0.1.0 is equivalent to uds deploy ghcr.io/defenseunicorns/packages/uds/bundles/unicorn-bundle:v0.1.0

The bundle matching and expansion is ordered as follows:

  1. Local with a tar.zst extension
  2. Remote path: oci://ghcr.io/defenseunicorns/packages/uds/bundles/<path>
  3. Remote path: oci://ghcr.io/defenseunicorns/packages/delivery/<path>
  4. Remote path: oci://ghcr.io/defenseunicorns/packages/<path>

That is to say, if the bundle is not local, UDS CLI will check path 2, path 3, etc for the remote bundle artifact. This behavior can be overridden by specifying the full path to the bundle artifact, for example uds deploy ghcr.io/defenseunicorns/dev/path/dev-bundle:v0.1.0.

Bundle Create

Pulls the Zarf packages from the registry and bundles them into an OCI artifact.

There are 2 ways to create Bundles:

  1. Inside an OCI registry: uds create <dir> -o ghcr.io/defenseunicorns/dev
  2. Locally on your filesystem: uds create <dir>

Bundle Deploy

Deploys the bundle

There are 2 ways to deploy Bundles:

  1. From an OCI registry: uds deploy ghcr.io/defenseunicorns/dev/<name>:<tag>
  2. From your local filesystem: uds deploy uds-bundle-<name>.tar.zst

Specifying Packages using --packages

By default all the packages in the bundle are deployed, but you can also deploy only certain packages in the bundle by using the --packages flag.

As an example: uds deploy uds-bundle-<name>.tar.zst --packages init,nginx

Resuming Bundle Deploys using --resume

By default all the packages in the bundle are deployed, regardless of if they have already been deployed, but you can also choose to only deploy packages that have not already been deployed by using the --resume flag

As an example: uds deploy uds-bundle-<name>.tar.zst --resume

Pruning Unreferenced Packages

In the process of upgrading bundles, it’s common to swap or remove packages from a uds-bundle.yaml. These packages can become unreferenced, meaning that they are still deployed to the cluster, but are no longer referenced by a bundle. To remove these packages from the cluster, you can use the --prune flag when deploying a bundle.

Pre-Deploy View

When uds deploy is executed, the bundle’s metadata, along with a list of its packages and each package’s overrides and Zarf variables, will be outputted to the terminal. Unlike inspect --list-variables, this output will show the value set for each override or Zarf variable. Overrides and variables that have not been set will not be shown in the output.

Bundle Inspect

Inspect the uds-bundle.yaml of a bundle

  1. From an OCI registry: uds inspect oci://ghcr.io/defenseunicorns/dev/<name>:<tag>
  2. From your local filesystem: uds inspect uds-bundle-<name>.tar.zst

Viewing Images in a Bundle

It is possible to derive images from a uds-bundle.yaml, local UDS tarball artifacts, and remote OCI repos. This can be useful for situations where you need to know what images will be bundled before you actually create the bundle or what images will be deployed if using an already created bundle. This is accomplished with the --list-images flag. For example:

uds inspect --list-images [BUNDLE_YAML_FILE|BUNDLE_TARBALL|OCI_REF]

This command will return a list of images derived from the bundle’s packages, taking into account optional and required package components.

The list of images will be grouped by package they are derived from and outputted in a YAML format.

e.g. uds inspect k3d-core-slim-dev:0.26.0 --list-images

core-slim-dev:
- docker.io/istio/pilot:1.22.3-distroless
- docker.io/istio/proxyv2:1.22.3-distroless
- ghcr.io/defenseunicorns/pepr/controller:v0.34.1
- quay.io/keycloak/keycloak:24.0.5
- ghcr.io/defenseunicorns/uds/identity-config:0.6.0
init:
- library/registry:2.8.3
- library/registry:2.8.3
- ghcr.io/zarf-dev/zarf/agent:v0.38.2

To extract only the image names and de-dupe:

uds inspect k3d-core-slim-dev:0.26.0 --list-images | yq '.[] | .[]' | sort | uniq

docker.io/istio/pilot:1.22.3-distroless
docker.io/istio/proxyv2:1.22.3-distroless
ghcr.io/defenseunicorns/pepr/controller:v0.34.1
ghcr.io/defenseunicorns/uds/identity-config:0.6.0
ghcr.io/zarf-dev/zarf/agent:v0.38.2
library/registry:2.8.3
quay.io/keycloak/keycloak:24.0.5

Viewing SBOMs

There are 2 additional flags for the uds inspect command you can use to extract and view SBOMs:

  • Output the SBOMs as a tar file: uds inspect ... --sbom
  • Output SBOMs into a directory as files: uds inspect ... --sbom --extract

This functionality will use the sboms.tar of the underlying Zarf packages to create new a bundle-sboms.tar artifact containing all SBOMs from the Zarf packages in the bundle.

Viewing Variables

To view the configurable overrides and Zarf variables of a bundle’s packages:

uds inspect --list-variables [BUNDLE_YAML_FILE|BUNDLE_TARBALL|OCI_REF]

Bundle Publish

Local bundles can be published to an OCI registry like so: uds publish <bundle>.tar.zst oci://<registry>

As an example: uds publish uds-bundle-example-arm64-0.0.1.tar.zst oci://ghcr.io/github_user

Tagging

Bundles, by default, are tagged based on the bundle version found in the metadata of the uds-bundle.yaml file. To override the default tag, you can use the --version flag like so:

uds publish uds-bundle-example-arm64-0.0.1.tar.zst oci://ghcr.io/github_user --version <custom-tag>

Bundle Remove

Removes the bundle

There are 2 ways to remove Bundles:

  1. From an OCI registry: uds remove oci://ghcr.io/defenseunicorns/dev/<name>:<tag> --confirm
  2. From your local filesystem: uds remove uds-bundle-<name>.tar.zst --confirm

By default all the packages in the bundle are removed, but you can also remove only certain packages in the bundle by using the --packages flag.

As an example: uds remove uds-bundle-<name>.tar.zst --packages init,nginx

Logs

The uds logs command can be used to view the most recent logs of a bundle operation. Note that depending on your OS temporary directory and file settings, recent logs are purged after a certain amount of time, so this command may return an error if the logs are no longer available.

Bundle Architecture and Multi-Arch Support

There are several ways to specify the architecture of a bundle according to the following precedence:

  1. Setting --architecture or -a flag during uds ... operations: uds create <dir> --architecture arm64
  2. Setting a UDS_ARCHITECTURE environment variable
  3. Setting the options.architecture key in a uds-config.yaml
  4. Setting the metadata.architecture key in a uds-bundle.yaml

This means that setting the --architecture flag takes precedence over all other methods of specifying the architecture.

UDS CLI supports multi-arch bundles. This means you can push bundles with different architectures to the same remote OCI repository, at the same tag. For example, you can push both an amd64 and arm64 bundle to ghcr.io/<org>/<bundle name>:0.0.1.

Architecture Validation

When deploying a local bundle, the bundle’s architecture will be used for comparison against the cluster architecture to ensure compatibility. If deploying a remote bundle, by default the bundle is pulled based on system architecture, which is then checked against the cluster.

If, for example, you have a multi-arch remote bundle that you want to deploy from an arm64 machine to an amd64 cluster, the validation with fail because the system arch does not match the cluster arch. However, you can pull the correct bundle version by specifying the arch with the command line architecture flag.

e.g. uds deploy -a amd64 <remote-multi-arch-bundle.tar.zst> --confirm

Variables and Configuration

The UDS CLI can be configured with a uds-config.yaml file. This file can be placed in the current working directory or specified with an environment variable called UDS_CONFIG. The basic structure of the uds-config.yaml is as follows:

options:
log_level: debug
architecture: arm64
no_log_file: false
no_progress: false
uds_cache: /tmp/uds-cache
tmp_dir: /tmp/tmp_dir
insecure: false
oci_concurrency: 3
shared:
domain: uds.dev # shared across all packages in a bundle
variables:
my-zarf-package: # name of Zarf package
ui_color: green # key is not case sensitive and refers to name of Zarf variable
UI_MSG: "Hello Unicorn"
hosts: # variables can be complex types such as lists and maps
- host: burning.boats
paths:
- path: "/"
pathType: "Prefix"

The options key contains UDS CLI options that are not specific to a particular Zarf package. The variables key contains variables that are specific to a particular Zarf package. If you want to share insensitive variables across multiple Zarf packages, you can use the shared key, where the key is the variable name and the value is the variable value.

Sharing Variables

Zarf package variables can be passed between Zarf packages:

kind: UDSBundle
metadata:
name: simple-vars
description: show how vars work
version: 0.0.1
packages:
- name: output-var
repository: localhost:888/output-var
ref: 0.0.1
exports:
- name: OUTPUT
- name: receive-var
repository: localhost:888/receive-var
ref: 0.0.1
imports:
- name: OUTPUT
package: output-var

Variables that you want to make available to other packages are in the export block of the Zarf package to export a variable from. By default, all exported variables are available to all of the packages in a bundle. To have another package ingest a specific exported variable, like in the case of variable name collisions, use the imports key to name both the variable and package that the variable is exported from, like in the example above.

In the example above, the OUTPUT variable is created as part of a Zarf Action in the output-var package, and the receive-var package expects a variable called OUTPUT.

Sharing Variables Across Multiple Packages

If a Zarf variable has the same name in multiple packages and you don’t want to set it multiple times via the import/export syntax, you can set an environment variable prefixed with UDS_ and it will be applied to all the Zarf packages in a bundle. For example, if multiple packages require a DOMAIN variable, you could set it once with a UDS_DOMAIN environment variable and it would be applied to all packages. Note that this can also be done with the shared key in the uds-config.yaml file.

On deploy, you can also set package variables by using the --set flag. If the package name isn’t included in the key (example: --set super=true) the variable will get applied to all of the packages. If the package name is included in the key (example: --set cool-package.super=true) the variable will only get applied to that package.

Variable Precedence and Specificity

In a bundle, variables can come from 6 sources. Those sources and their precedence are shown below in order of least to most specificity:

  • declared in a Zarf pkg
  • import’ed from a bundle package’s export
  • configured in the shared key in a uds-config.yaml
  • configured in the variables key in a uds-config.yaml
  • set with an environment variable prefixed with UDS_ (ex. UDS_OUTPUT)
  • set using the --set flag when running the uds deploy command

That is to say, variables set using the --set flag take precedence over all other variable sources.

Configuring Zarf Init Packages

Zarf init packages that are typically deployed using zarf init have a few special flags that are attached to that command. These options can be configured like any other variable: specified in a uds-config.yaml, as an environment variable prefixed with UDS_ or via the --set flag.

uds-config.yaml
variables:
zarf-init:
INIT_REGISTRY_URL: "https://my-registry.io"
INIT_REGISTRY_PUSH_USERNAME: "registry-user"

Duplicate Packages And Naming

It is possible to deploy multiple instances of the same Zarf package in a bundle. For example, the following uds-bundle.yaml deploys 3 instances of the helm-overrides Zarf packages:

kind: UDSBundle
metadata:
name: duplicates
description: testing a bundle with duplicate packages in specified namespaces
version: 0.0.1
packages:
- name: helm-overrides
repository: localhost:5000/helm-overrides
ref: 0.0.1
overrides:
podinfo-component:
unicorn-podinfo: # name of Helm chart
namespace: podinfo-ns
# note the unique name and namespace
- name: helm-overrides-duplicate
repository: localhost:5000/helm-overrides
ref: 0.0.1
overrides:
podinfo-component:
unicorn-podinfo:
namespace: another-podinfo-ns
# note the unique name, namespace and the path to the Zarf package tarball
- name: helm-overrides-local-duplicate
path: src/test/packages/helm/zarf-package-helm-overrides-arm64-0.0.1.tar.zst
ref: 0.0.1
overrides:
podinfo-component:
unicorn-podinfo:
namespace: yet-another-podinfo-ns

The naming conventions for deploying duplicate packages are as follows:

  1. The name field of the package in the uds-bundle.yaml must be unique
  2. The duplicate packages must be deployed in different namespaces
  3. In order to deploy duplicates of local packages, the path field must point to a Zarf package tarball instead of to a folder.

Zarf Integration

UDS CLI includes a vendored version of Zarf inside of its binary. To use Zarf, simply run uds zarf <command>. For example, to create a Zarf package, run uds zarf create <dir>, or to use the airgap tooling that Zarf provides, run uds zarf tools <cmd>.

Dev Mode

Dev mode facilitates faster dev cycles when developing and testing bundles

Terminal window
uds dev deploy <path-to-bundle-yaml-dir> | <oci-ref>

The dev deploy command performs the following operations:

  • Deploys the bundle in YOLO mode, eliminating the need to do a zarf init
    • Any kind: ZarfInitConfig packages in the bundle will be ignored
  • For local bundles:
    • For local packages:
      • Creates the Zarf tarball if one does not already exist or the --force-create flag can be used to force the creation of a new Zarf package
        • The Zarf tarball is created in the same directory as the zarf.yaml
        • The --flavor flag can be used to specify what flavor of a package you want to create (example: --flavor podinfo=upstream to specify the flavor for the podinfo package or --flavor upstream to specify the flavor for all the packages in the bundle)
    • For remote packages:
      • The --ref flag can be used to specify what package ref you want to deploy (example: --ref podinfo=0.2.0)
    • Creates a bundle from the newly created Zarf packages

Monitor

UDS CLI provides a uds monitor command that can be used to monitor the status of a UDS cluster

Monitor Pepr

To monitor the status of a UDS cluster’s admission and operator controllers, run: uds monitor pepr

UDS Controllers

UDS clusters contain two Kubernetes controllers, both created using Pepr:

  1. Admission Controller: Corresponds to the pepr-uds-core pods in the cluster. This controller is responsible for validating and mutating resources in the cluster including the enforcement of UDS Exemptions.

  2. Operator Controller: Corresponds to the pepr-uds-core-watcher pods. This controller is responsible for managing the lifecyle of UDS Package resources in the cluster.

Monitor Args

Aggregate all admission and operator logs into a single stream:

Terminal window
uds monitor pepr

Stream UDS Operator actions (UDS Package processing, status updates, and errors):

Terminal window
uds monitor pepr operator

Stream UDS Policy logs (Allow, Deny, Mutate):

Terminal window
uds monitor pepr policies

Stream UDS Policy allow logs:

Terminal window
uds monitor pepr allowed

Stream UDS Policy deny logs:

Terminal window
uds monitor pepr denied

Stream UDS Policy mutation logs:

Terminal window
uds monitor pepr mutated

Stream UDS Policy deny logs and UDS Operator error logs: uds monitor pepr failed

Monitor Flags

-f, --follow Continuously stream Pepr logs

--json Return the raw JSON output of the logs “ `—since duration Only return logs newer than a relative duration like 5s, 2m, or 3h. Defaults to all logs.

-t, --timestamps Show timestamps in Pepr log

Scan

The scan command is used to scan a Zarf package for vulnerabilities and generate a report. This command is currently in ALPHA.

Usage

To use the scan command, run:

Terminal window
uds scan --org <organization> --package-name <package-name> --tag <tag> [options]

Required Parameters

  • --org or -o: Organization name (default: defenseunicorns)
  • --package-name or -n: Name of the package (e.g., packages/uds/gitlab-runner)
  • --tag or -g: Tag name (e.g., 16.10.0-uds.0-upstream)
  • --output-file or -f: Output file for CSV results

Optional Parameters

  • --docker-username or -u: Docker username for registry access, accepts CSV values
  • --docker-password or -p: Docker password for registry access, accepts CSV values

Example Usage

Terminal window
uds scan -o defenseunicorns -n packages/uds/gitlab-runner -g 16.10.0-uds.0-upstream -u docker-username -p docker-password -f gitlab-runner.csv