GoogleContainerTools / kaniko – no longer a maintained projec

The code remains available for historic purposes.

The README as of the archival date remains unchanged below for historic purposes.

kaniko is a tool to build container images from a Dockerfile, inside a container or Kubernetes cluster.

kaniko doesn't depend on a Docker daemon and executes each command within a Dockerfile completely in userspace. This enables building container images in environments that can't easily or securely run a Docker daemon, such as a standard Kubernetes cluster.

kaniko is meant to be run as an image: gcr.io/kaniko-project/executor. We do not recommend running the kaniko executor binary in another image, as it might not work as you expect - see Known Issues.

We'd love to hear from you! Join us on #kaniko Kubernetes Slack

📣 Please fill out our quick 5-question survey so that we can learn how satisfied you are with kaniko, and what improvements we should make. Thank you! 👯

If you are interested in contributing to kaniko, see DEVELOPMENT.md and CONTRIBUTING.md.

Table of Contents generated with DocToc

• kaniko - Build Images In Kubernetes
  • 🚨NOTE: kaniko is not an officially supported Google product🚨
  • Community
  • How does kaniko work?
  • Known Issues
  • Demo
  • Tutorial
  • Using kaniko
    • kaniko Build Contexts
    • Using Azure Blob Storage
    • Using Private Git Repository
    • Using Standard Input
    • Running kaniko
      • Running kaniko in a Kubernetes cluster
        • Kubernetes secret
      • Running kaniko in gVisor
      • Running kaniko in Google Cloud Build
      • Running kaniko in Docker
    • Caching
      • Caching Layers
      • Caching Base Images
    • Pushing to Different Registries
      • Pushing to Docker Hub
      • Pushing to Google GCR
      • Pushing to GCR using Workload Identity
      • Pushing to Amazon ECR
      • Pushing to Azure Container Registry
      • Pushing to JFrog Container Registry or to JFrog Artifactory
    • Additional Flags
      • Flag --build-arg
      • Flag --cache
      • Flag --cache-dir
      • Flag --cache-repo
      • Flag --cache-copy-layers
      • Flag --cache-run-layers
      • Flag --cache-ttl duration
      • Flag --cleanup
      • Flag --compressed-caching
      • Flag --context-sub-path
      • Flag --custom-platform
      • Flag --digest-file
      • Flag --dockerfile
      • Flag --force
      • Flag --git
      • Flag --image-name-with-digest-file
      • Flag --image-name-tag-with-digest-file
      • Flag --insecure
      • Flag --insecure-pull
      • Flag --insecure-registry
      • Flag --label
      • Flag --log-format
      • Flag --log-timestamp
      • Flag --no-push
      • Flag --no-push-cache
      • Flag --oci-layout-path
      • Flag --push-retry
      • Flag --registry-certificate
      • Flag --registry-client-cert
      • Flag --registry-map
      • Flag --registry-mirror
      • Flag --skip-default-registry-fallback
      • Flag --reproducible
      • Flag --single-snapshot
      • Flag --skip-push-permission-check
      • Flag --skip-tls-verify
      • Flag --skip-tls-verify-pull
      • Flag --skip-tls-verify-registry
      • Flag --skip-unused-stages
      • Flag --snapshot-mode
      • Flag --tar-path
      • Flag --target
      • Flag --use-new-run
      • Flag --verbosity
      • Flag --ignore-var-run
      • Flag --ignore-path
      • Flag --image-fs-extract-retry
      • Flag --image-download-retry
    • Debug Image
  • Security
    • Verifying Signed Kaniko Images
  • Kaniko Builds - Profiling
  • Creating Multi-arch Container Manifests Using Kaniko and Manifest-tool
    • General Workflow
    • Limitations and Pitfalls
    • Example CI Pipeline (GitLab)
      • Building the Separate Container Images
      • Merging the Container Manifests
      • On the Note of Adding Versioned Tags
  • Comparison with Other Tools
  • Community
  • Limitations
    • mtime and snapshotting
  • References

We'd love to hear from you! Join #kaniko on Kubernetes Slack

The kaniko executor image is responsible for building an image from a Dockerfile and pushing it to a registry. Within the executor image, we extract the filesystem of the base image (the FROM image in the Dockerfile). We then execute the commands in the Dockerfile, snapshotting the filesystem in userspace after each one. After each command, we append a layer of changed files to the base image (if there are any) and update image metadata.

• kaniko does not support building Windows containers.
• Running kaniko in any Docker image other than the official kaniko image is not supported due to implementation details.
  • This includes copying the kaniko executables from the official image into another image (e.g. a Jenkins CI agent).
  • In particular, it cannot use chroot or bind-mount because its container must not require privilege, so it unpacks directly into its own container root and may overwrite anything already there.
• kaniko does not support the v1 Registry API (Registry v1 API Deprecation)

For a detailed example of kaniko with local storage, please refer to a getting started tutorial.

Please see References for more docs & video tutorials

To use kaniko to build and push an image for you, you will need:

1. A build context, aka something to build
2. A running instance of kaniko

kaniko's build context is very similar to the build context you would send your Docker daemon for an image build; it represents a directory containing a Dockerfile which kaniko will use to build your image. For example, a COPY command in your Dockerfile should refer to a file in the build context.

You will need to store your build context in a place that kaniko can access. Right now, kaniko supports these storage solutions:

• GCS Bucket
• S3 Bucket
• Azure Blob Storage
• Local Directory
• Local Tar
• Standard Input
• Git Repository

Note about Local Directory: this option refers to a directory within the kaniko container. If you wish to use this option, you will need to mount in your build context into the container as a directory.

Note about Local Tar: this option refers to a tar gz file within the kaniko container. If you wish to use this option, you will need to mount in your build context into the container as a file.

Note about Standard Input: the only Standard Input allowed by kaniko is in .tar.gz format.

If using a GCS or S3 bucket, you will first need to create a compressed tar of your build context and upload it to your bucket. Once running, kaniko will then download and unpack the compressed tar of the build context before starting the image build.

To create a compressed tar, you can run:

Then, copy over the compressed tar into your bucket. For example, we can copy over the compressed tar to a GCS bucket with gsutil:

When running kaniko, use the --context flag with the appropriate prefix to specify the location of your build context:

If you don't specify a prefix, kaniko will assume a local directory. For example, to use a GCS bucket called kaniko-bucket, you would pass in --context=gs://kaniko-bucket/path/to/context.tar.gz.

If you are using Azure Blob Storage for context file, you will need to pass Azure Storage Account Access Key as an environment variable named AZURE_STORAGE_ACCESS_KEY through Kubernetes Secrets

You can use Personal Access Tokens for Build Contexts from Private Repositories from GitHub.

You can either pass this in as part of the git URL (e.g., git://[email protected]/acme/myproject.git#refs/heads/mybranch) or using the environment variable GIT_TOKEN.

You can also pass GIT_USERNAME and GIT_PASSWORD (password being the token) if you want to be explicit about the username.

If running kaniko and using Standard Input build context, you will need to add the docker or kubernetes -i, --interactive flag. Once running, kaniko will then get the data from STDIN and create the build context as a compressed tar. It will then unpack the compressed tar of the build context before starting the image build. If no data is piped during the interactive run, you will need to send the EOF signal by yourself by pressing Ctrl+D.

Complete example of how to interactively run kaniko with .tar.gz Standard Input data, using docker:

Complete example of how to interactively run kaniko with .tar.gz Standard Input data, using Kubernetes command line with a temporary container and completely dockerless:

There are several different ways to deploy and run kaniko:

• In a Kubernetes cluster
• In gVisor
• In Google Cloud Build
• In Docker

Requirements:

• Standard Kubernetes cluster (e.g. using GKE)
• Kubernetes Secret
• A build context

To run kaniko in a Kubernetes cluster, you will need a standard running Kubernetes cluster and a Kubernetes secret, which contains the auth required to push the final image.

To create a secret to authenticate to Google Cloud Registry, follow these steps:

1. Create a service account in the Google Cloud Console project you want to push the final image to with Storage Admin permissions.
2. Download a JSON key for this service account
3. Rename the key to kaniko-secret.json
4. To create the secret, run:

Note: If using a GCS bucket in the same GCP project as a build context, this service account should now also have permissions to read from that bucket.

The Kubernetes Pod spec should look similar to this, with the args parameters filled in:

This example pulls the build context from a GCS bucket. To use a local directory build context, you could consider using configMaps to mount in small build contexts.

Running kaniko in gVisor provides an additional security boundary. You will need to add the --force flag to run kaniko in gVisor, since currently there isn't a way to determine whether or not a container is running in gVisor.

We pass in --runtime=runsc to use gVisor. This example mounts the current directory to /workspace for the build context and the ~/.config directory for GCR credentials.

Requirements:

• A build context

To run kaniko in GCB, add it to your build config as a build step:

kaniko will build and push the final image in this build step.

Requirements:

• Docker

We can run the kaniko executor image locally in a Docker daemon to build and push an image from a Dockerfile.

For example, when using gcloud and GCR you could run kaniko as follows:

There is also a utility script run_in_docker.sh that can be used as follows:

NOTE: run_in_docker.sh expects a path to a Dockerfile relative to the absolute path of the build context.

An example run, specifying the Dockerfile in the container directory /workspace, the build context in the local directory /home/user/kaniko-project, and a Google Container Registry as a remote image destination:

kaniko can cache layers created by RUN(configured by flag --cache-run-layers) and COPY (configured by flag --cache-copy-layers) commands in a remote repository. Before executing a command, kaniko checks the cache for the layer. If it exists, kaniko will pull and extract the cached layer instead of executing the command. If not, kaniko will execute the command and then push the newly created layer to the cache.

Note that kaniko cannot read layers from the cache after a cache miss: once a layer has not been found in the cache, all subsequent layers are built locally without consulting the cache.

Users can opt into caching by setting the --cache=true flag. A remote repository for storing cached layers can be provided via the --cache-repo flag. If this flag isn't provided, a cached repo will be inferred from the --destination provided.

kaniko can cache images in a local directory that can be volume mounted into the kaniko pod. To do so, the cache must first be populated, as it is read-only. We provide a kaniko cache warming image at gcr.io/kaniko-project/warmer:

--image can be specified for any number of desired images. --dockerfile can be specified for the path of dockerfile for cache.These command will combined to cache those images by digest in a local directory named cache. Once the cache is populated, caching is opted into with the same --cache=true flag as above. The location of the local cache is provided via the --cache-dir flag, defaulting to /cache as with the cache warmer. See the examples directory for how to use with kubernetes clusters and persistent cache volumes.

kaniko uses Docker credential helpers to push images to a registry.

kaniko comes with support for GCR, Docker config.json and Amazon ECR, but configuring another credential helper should allow pushing to a different registry.

Get your docker registry user and password encoded in base64

Create a config.json file with your Docker registry url and the previous generated base64 string

Note: Please use v1 endpoint. See #1209 for more details

Run kaniko with the config.json inside /kaniko/.docker/config.json

To create a credentials to authenticate to Google Cloud Registry, follow these steps:

1. Create a service account or in the Google Cloud Console project you want to push the final image to with Storage Admin permissions.
2. Download a JSON key for this service account
3. (optional) Rename the key to kaniko-secret.json, if you don't rename, you have to change the name used the command(in the volume part)
4. Run the container adding the path in GOOGLE_APPLICATION_CREDENTIALS env var

If you have enabled Workload Identity on your GKE cluster then you can use the workload identity to push built images to GCR without adding a GOOGLE_APPLICATION_CREDENTIALS in your kaniko pod specification.

Learn more on how to enable and migrate existing apps to workload identity.

To authenticate using workload identity you need to run the kaniko pod using the Kubernetes Service Account (KSA) bound to Google Service Account (GSA) which has Storage.Admin permissions to push images to Google Container registry.

Please follow the detailed steps here to create a Kubernetes Service Account, Google Service Account and create an IAM policy binding between the two to allow the Kubernetes Service account to act as the Google service account.

To grant the Google Service account the right permission to push to GCR, run the following GCR command

Please ensure, kaniko pod is running in the namespace and with a Kubernetes Service Account.

The Amazon ECR credential helper is built into the kaniko executor image.

1. Configure credentials

   1. You can use instance roles when pushing to ECR from a EC2 instance or from EKS, by configuring the instance role permissions (the AWS managed policy EC2InstanceProfileForImageBuilderECRContainerBuilds provides broad permissions to upload ECR images and may be used as configuration baseline). Additionally, set AWS_SDK_LOAD_CONFIG=true as environment variable within the kaniko pod. If running on an EC2 instance with an instance profile, you may also need to set AWS_EC2_METADATA_DISABLED=true for kaniko to pick up the correct credentials.

   2. Or you can create a Kubernetes secret for your ~/.aws/credentials file so that credentials can be accessed within the cluster. To create the secret, run: shell kubectl create secret generic aws-secret --from-file=<path to .aws/credentials>

The Kubernetes Pod spec should look similar to this, with the args parameters filled in. Note that aws-secret volume mount and volume are only needed when using AWS credentials from a secret, not when using instance roles.

An ACR credential helper is built into the kaniko executor image, which can be used to authenticate with well-known Azure environmental information.

To configure credentials, you will need to do the following:

1. Update the credStore section of config.json:

A downside of this approach is that ACR authentication will be used for all registries, which will fail if you also pull from DockerHub, GCR, etc. Thus, it is better to configure the credential tool only for your ACR registries by using credHelpers instead of credsStore:

You can mount in the new config as a configMap:

2. Configure credentials

You can create a Kubernetes secret with environment variables required for Service Principal authentication and expose them to the builder container.

If the above are not set then authentication falls back to managed service identities and the MSI endpoint is attempted to be contacted which will work in various Azure contexts such as App Service and Azure Kubernetes Service where the MSI endpoint will authenticate the MSI context the service is running under.

The Kubernetes Pod spec should look similar to this, with the args parameters filled in. Note that azure-secret secret is only needed when using Azure Service Principal credentials, not when using a managed service identity.

Kaniko can be used with both JFrog Container Registry and JFrog Artifactory.

Get your JFrog Artifactory registry user and password encoded in base64

Create a config.json file with your Artifactory Docker local registry URL and the previous generated base64 string

For example, for Artifactory cloud users, the docker registry should be: <company>.<local-repository-name>.io.

Run kaniko with the config.json inside /kaniko/.docker/config.json

After the image is uploaded, using the JFrog CLI, you can collect and publish the build information to Artifactory and trigger build vulnerabilities scanning using JFrog Xray.

To collect and publish the image's build information using the Jenkins Artifactory plugin, see instructions for scripted pipeline and declarative pipeline.

This flag allows you to pass in ARG values at build time, similarly to Docker. You can set it multiple times for multiple arguments.

Note that passing values that contain spaces is not natively supported - you need to ensure that the IFS is set to null before your executor command. You can set this by adding export IFS='' before your executor call. See the following example

Set this flag as --cache=true to opt into caching with kaniko.

Set this flag to specify a local directory cache for base images. Defaults to /cache.

This flag must be used in conjunction with the --cache=true flag.

Set this flag to specify a remote repository that will be used to store cached layers.

If this flag is not provided, a cache repo will be inferred from the --destination flag. If --destination=gcr.io/kaniko-project/test, then cached layers will be stored in gcr.io/kaniko-project/test/cache.

This flag must be used in conjunction with the --cache=true flag.

Set this flag to cache copy layers.

Set this flag to cache run layers (default=true).

Cache timeout in hours. Defaults to two weeks.

Set this flag to clean the filesystem at the end of the build.

Set this to false in order to prevent tar compression for cached layers. This will increase the runtime of the build, but decrease the memory usage especially for large builds. Try to use --compressed-caching=false if your build fails with an out of memory error. Defaults to true.

Set a sub path within the given --context.

Its particularly useful when your context is, for example, a git repository, and you want to build one of its subfolders instead of the root folder.

Allows to build with another default platform than the host, similarly to docker build --platform xxx the value has to be on the form --custom-platform=linux/arm, with acceptable values listed here: GOOS/GOARCH.

It's also possible specifying CPU variants adding it as a third parameter (like --custom-platform=linux/arm/v5). Currently CPU variants are only known to be used for the ARM architecture as listed here: GOARM

The resulting images cannot provide any metadata about CPU variant due to a limitation of the OCI-image specification.

This is not virtualization and cannot help to build an architecture not natively supported by the build host. This is used to build i386 on an amd64 Host for example, or arm32 on an arm64 host.

Set this flag to specify a file in the container. This file will receive the digest of a built image. This can be used to automatically track the exact image built by kaniko.

For example, setting the flag to --digest-file=/dev/termination-log will write the digest to that file, which is picked up by Kubernetes automatically as the {{.state.terminated.message}} of the container.

Path to the dockerfile to be built. (default "Dockerfile")

Force building outside of a container

Branch to clone if build context is a git repository (default branch=,single-branch=false,recurse-submodules=false,insecure-skip-tls=false)

Specify a file to save the image name w/ digest of the built image to.

Specify a file to save the image name w/ image tag and digest of the built image to.

Set this flag if you want to push images to a plain HTTP registry. It is supposed to be used for testing purposes only and should not be used in production!

Set this flag if you want to pull images from a plain HTTP registry. It is supposed to be used for testing purposes only and should not be used in production!

You can set --insecure-registry <registry-name> to use plain HTTP requests when accessing the specified registry. It is supposed to be used for testing purposes only and should not be used in production! You can set it multiple times for multiple registries.

Set this flag as --label key=value to set some metadata to the final image. This is equivalent as using the LABEL within the Dockerfile.

Set this flag as --log-format=<text|color|json> to set the log format. Defaults to color.

Set this flag as --log-timestamp=<true|false> to add timestamps to <text|color> log format. Defaults to false.

Set this flag if you only want to build the image, without pushing to a registry. This can also be defined through KANIKO_NO_PUSH environment variable.

NOTE: this will still push cache layers to the repo, to disable pushing cache layers use --no-push-cache

Set this flag if you do not want to push cache layers to a registry. Can be used in addition to --no-push to push no layers to a registry.

Set this flag to specify a directory in the container where the OCI image layout of a built image will be placed. This can be used to automatically track the exact image built by kaniko.

For example, to surface the image digest built in a Tekton task, this flag should be set to match the image resource outputImageDir.

Note: Depending on the built image, the media type of the image manifest might be either application/vnd.oci.image.manifest.v1+json or application/vnd.docker.distribution.manifest.v2+json.

Set this boolean flag to true if you want the Kaniko process to exit with success when a push error related to tag immutability occurs.

This is useful for example if you have parallel builds pushing the same tag and do not care which one actually succeeds.

Defaults to false.

Set this flag to the number of retries that should happen for the push of an image to a remote destination. Defaults to 0.

Set this flag to provide a certificate for TLS communication with a given registry.

Expected format is my.registry.url=/path/to/the/certificate.cert

Set this flag to provide a certificate/key pair for mutual TLS (mTLS) communication with a given registry that requires mTLS for authentication.

Expected format is my.registry.url=/path/to/client/cert.crt,/path/to/client/key.key

Set this flag if you want to remap registries references. Usefull for air gap environement for example. You can use this flag more than once, if you want to set multiple mirrors for a given registry. You can mention several remap in a single flag too, separated by semi-colon. If an image is not found on the first mirror, Kaniko will try the next mirror(s), and at the end fallback on the original registry.

Registry maps can also be defined through KANIKO_REGISTRY_MAP environment variable.

Expected format is original-registry=remapped-registry[;another-reg=another-remap[;...]] for example.

Note that you can specify a URL with scheme for this flag. Some valid options are:

• index.docker.io=mirror.gcr.io
• gcr.io=127.0.0.1
• quay.io=192.168.0.1:5000
• index.docker.io=docker-io.mirrors.corp.net;index.docker.io=mirror.gcr.io;gcr.io=127.0.0.1 will try docker-io.mirrors.corp.net then mirror.gcr.io for index.docker.io and 127.0.0.1 for gcr.io
• docker.io=harbor.provate.io/theproject

Set this flag if you want to use a registry mirror instead of the default index.docker.io. You can use this flag more than once, if you want to set multiple mirrors. If an image is not found on the first mirror, Kaniko will try the next mirror(s), and at the end fallback on the default registry.

Mirror can also be defined through KANIKO_REGISTRY_MIRROR environment variable.

Expected format is mirror.gcr.io or mirror.gcr.io/path for example.

Note that you can specify a URL with scheme for this flag. Some valid options are:

• mirror.gcr.io
• 127.0.0.1
• 192.168.0.1:5000
• mycompany-docker-virtual.jfrog.io
• harbor.provate.io/theproject

Set this flag if you want the build process to fail if none of the mirrors listed in flag registry-mirror can pull some image. This should be used with mirrors that implements a whitelist or some image restrictions.

If registry-mirror is not set or is empty, this flag is ignored.

Set this flag to strip timestamps out of the built image and make it reproducible.

This flag takes a single snapshot of the filesystem at the end of the build, so only one layer will be appended to the base image.

Set this flag to skip push permission check. This can be useful to delay Kanikos first request for delayed network-policies.

Set this flag to skip TLS certificate validation when pushing to a registry. It is supposed to be used for testing purposes only and should not be used in production!

Set this flag to skip TLS certificate validation when pulling from a registry. It is supposed to be used for testing purposes only and should not be used in production!

You can set --skip-tls-verify-registry <registry-name> to skip TLS certificate validation when accessing the specified registry. It is supposed to be used for testing purposes only and should not be used in production! You can set it multiple times for multiple registries.

This flag builds only used stages if defined to true. Otherwise it builds by default all stages, even the unnecessary ones until it reaches the target stage / end of Dockerfile

You can set the --snapshot-mode=<full (default), redo, time> flag to set how kaniko will snapshot the filesystem.

• If --snapshot-mode=full is set, the full file contents and metadata are considered when snapshotting. This is the least performant option, but also the most robust.

• If --snapshot-mode=redo is set, the file mtime, size, mode, owner uid and gid will be considered when snapshotting. This may be up to 50% faster than "full", particularly if your project has a large number files.

• If --snapshot-mode=time is set, only file mtime will be considered when snapshotting (see limitations related to mtime).

Set this flag as --tar-path=<path> to save the image as a tarball at path. You need to set --destination as well (for example --destination=image). If you want to save the image as tarball only you also need to set --no-push.

Set this flag to indicate which build stage is the target build stage.

Using this flag enables an experimental implementation of the Run command which does not rely on snapshotting at all. In this approach, in order to compute which files were changed, a marker file is created before executing the Run command. Then the entire filesystem is walked (takes ~1-3 seconds for 700Kfiles) to find all files whose ModTime is greater than the marker file. With this new run command implementation, the total build time is reduced seeing performance improvements in the range of ~75%. This new run mode trades off accuracy/correctness in some cases (potential for missed files in a "snapshot") for improved performance by avoiding the full filesystem snapshots.

Set this flag as --verbosity=<panic|fatal|error|warn|info|debug|trace> to set the logging level. Defaults to info.

Ignore /var/run when taking image snapshot. Set it to false to preserve /var/run/* in destination image. (Default true).

Set this flag as --ignore-path=<path> to ignore path when taking an image snapshot. Set it multiple times for multiple ignore paths.

Set this flag to the number of retries that should happen for the extracting an image filesystem. Defaults to 0.

Set this flag to the number of retries that should happen when downloading the remote image. Consecutive retries occur with exponential backoff and an initial delay of 1 second. Defaults to 0`.

The kaniko executor image is based on scratch and doesn't contain a shell. We provide gcr.io/kaniko-project/executor:debug, a debug image which consists of the kaniko executor image along with a busybox shell to enter.

You can launch the debug image with a shell entrypoint:

kaniko by itself does not make it safe to run untrusted builds inside your cluster, or anywhere else.

kaniko relies on the security features of your container runtime to provide build security.

The minimum permissions kaniko needs inside your container are governed by a few things:

• The permissions required to unpack your base image into its container
• The permissions required to execute the RUN commands inside the container

If you have a minimal base image (SCRATCH or similar) that doesn't require permissions to unpack, and your Dockerfile doesn't execute any commands as the root user, you can run kaniko without root permissions. It should be noted that Docker runs as root by default, so you still require (in a sense) privileges to use kaniko.

You may be able to achieve the same default seccomp profile that Docker uses in your Pod by setting seccomp profiles with annotations on a PodSecurityPolicy to create or update security policies on your cluster.

kaniko images are signed for versions >= 1.5.2 using cosign!

To verify a public image, install cosign and use the provided public key:

If your builds are taking long, we recently added support to analyze kaniko function calls using Slow Jam To start profiling,

1. Add an environment variable STACKLOG_PATH to your pod definition.
2. If you are using the kaniko debug image, you can copy the file in the pre-stop container lifecycle hook.

While Kaniko itself currently does not support creating multi-arch manifests (contributions welcome), one can use tools such as manifest-tool to stitch multiple separate builds together into a single container manifest.

The general workflow for creating multi-arch manifests is as follows:

1. Build separate container images using Kaniko on build hosts matching your target architecture and tag them with the appropriate ARCH tag.
2. Push the separate images to your container registry.
3. Manifest-tool identifies the separate manifests in your container registry, according to a given template.
4. Manifest-tool pushes a combined manifest referencing the separate manifests.

The following conditions must be met:

1. You need access to build-machines running the desired architectures (running Kaniko in an emulator, e.g. QEMU should also be possible but goes beyond the scope of this documentation). This is something to keep in mind when using SaaS build tools such as github.com or gitlab.com, of which at the time of writing neither supports any non-x86_64 SaaS runners (GitHub,GitLab), so be prepared to bring your own machines (GitHub,GitLab.
2. Kaniko needs to be able to run on the desired architectures. At the time of writing, the official Kaniko container supports linux/amd64, linux/arm64, linux/s390x and linux/ppc64le (not on *-debug images).
3. The container registry of your choice must be OCIv1 or Docker v2.2 compatible.

It is up to you to find an automation tool that suits your needs best. We recommend using a modern CI/CD system such as GitHub workflows or GitLab CI. As we (the authors) happen to use GitLab CI, the following examples are tailored to this specific platform but the underlying principles should apply anywhere else and the examples are kept simple enough, so that you should be able to follow along, even without any previous experiences with this specific platform. When in doubt, visit the gitlab-ci.yml reference page for a comprehensive overview of the GitLab CI keywords.

gitlab-ci.yml:

gitlab-ci.yml:

For simplicity's sake we deliberately refrained from using versioned tagged images (all builds will be tagged as "latest") in the previous examples, as we feel like this adds to much platform and workflow specific code.

Nethertheless, for anyone interested in how we handle (dynamic) versioning in GitLab, here is a short rundown:

• If you are only interested in building tagged releases, you can simply use the GitLab predefined CI_COMMIT_TAG variable when running a tag pipeline.
• When you (like us) want to additionally build container images outside of releases, things get a bit messier. In our case, we added a additional job which runs before the build and merge jobs (don't forget to extend the needs section of the build and merge jobs accordingly), which will set the tag to latest when running on the default branch, to the commit hash when run on other branches and to the release tag when run on a tag pipeline.

gitlab-ci.yml:

Similar tools include:

• BuildKit
• img
• orca-build
• umoci
• buildah
• FTL
• Bazel rules_docker

All of these tools build container images with different approaches.

BuildKit (and img) can perform as a non-root user from within a container but requires seccomp and AppArmor to be disabled to create nested containers. kaniko does not actually create nested containers, so it does not require seccomp and AppArmor to be disabled. BuildKit supports "cross-building" multi-arch containers by leveraging QEMU.

orca-build depends on runc to build images from Dockerfiles, which can not run inside a container (for similar reasons to img above). kaniko doesn't use runc so it doesn't require the use of kernel namespacing techniques. However, orca-build does not require Docker or any privileged daemon (so builds can be done entirely without privilege).

umoci works without any privileges, and also has no restrictions on the root filesystem being extracted (though it requires additional handling if your filesystem is sufficiently complicated). However, it has no Dockerfile-like build tooling (it's a slightly lower-level tool that can be used to build such builders -- such as orca-build).

Buildah specializes in building OCI images. Buildah's commands replicate all of the commands that are found in a Dockerfile. This allows building images with and without Dockerfiles while not requiring any root privileges. Buildah’s ultimate goal is to provide a lower-level coreutils interface to build images. The flexibility of building images without Dockerfiles allows for the integration of other scripting languages into the build process. Buildah follows a simple fork-exec model and does not run as a daemon but it is based on a comprehensive API in golang, which can be vendored into other tools.

FTL and Bazel aim to achieve the fastest possible creation of Docker images for a subset of images. These can be thought of as a special-case "fast path" that can be used in conjunction with the support for general Dockerfiles kaniko provides.

kaniko-users Google group

To Contribute to kaniko, see DEVELOPMENT.md and CONTRIBUTING.md.

When taking a snapshot, kaniko's hashing algorithms include (or in the case of --snapshot-mode=time, only use) a file's mtime to determine if the file has changed. Unfortunately, there is a delay between when changes to a file are made and when the mtime is updated. This means:

• With the time-only snapshot mode (--snapshot-mode=time), kaniko may miss changes introduced by RUN commands entirely.
• With the default snapshot mode (--snapshot-mode=full), whether or not kaniko will add a layer in the case where a RUN command modifies a file but the contents do not change is theoretically non-deterministic. This does not affect the contents which will still be correct, but it does affect the number of layers.

Note that these issues are currently theoretical only. If you see this issue occur, please open an issue.

Kaniko currently supports COPY --chown and ADD --chown Dockerfile command. It does not support RUN --chown.

• Kaniko - Building Container Images In Kubernetes Without Docker.