Process
In general, the process to build a multi-cpu architecture container image follows the flow depicted on the following figure.
The commands to needed to implement the flow are described, using an example, in the next section.
Example
Following are the commands you need to create a multi-cpu architecture container image. Let's call the image onap/multi-cpu-app-py
Note that this flow can be used during the ONAP development-test-debug process. For the release process, the flow is implemented using CI/CD pipelines as shown in the next section.
Build arm image (A)
docker build -t onap/app-py-arm-linux .
Push arm image to the registry
docker push onap/app-py-arm-linux:latest
Build Intel image (B)
docker build -t onap/app-py-amd-linux .
Push Intel image to the registry
docker push onap/app-py-amd-linux:latest
Create the manifest list for image A and image B
docker manifest create onap/multi-cpu-app-py \ onap/app-py-arm-linux \ onap/app-py-amd-linux
Inspect the multi-cpu architecture manifest
docker manifest inspect --verbose onap/multi-cpu-app-py
Notice how the the manifest shows layers for both arm and Intel cpu architectures.
[ { "Ref": "docker.io/onap/app-py-amd-linux:latest", "Descriptor": { "mediaType": "application/vnd.docker.distribution.manifest.v2+json", "digest": "sha256:4b1b8361d47770668ff4c728dc388f376158d9e8d82c0b4cdb22147cc8ca1be6", "size": 1579, "platform": { "architecture": "amd64", "os": "linux" } }, "SchemaV2Manifest": { "schemaVersion": 2, "mediaType": "application/vnd.docker.distribution.manifest.v2+json", "config": { "mediaType": "application/vnd.docker.container.image.v1+json", "size": 6616, "digest": "sha256:f2947f4a0e9d5d0a4ccf74a7b3ad94611a8921ca022b154cabcad9e453ea6ec5" }, "layers": [ { "mediaType": "application/vnd.docker.image.rootfs.diff.tar.gzip", "size": 2206931, "digest": "sha256:4fe2ade4980c2dda4fc95858ebb981489baec8c1e4bd282ab1c3560be8ff9bde" }, { "mediaType": "application/vnd.docker.image.rootfs.diff.tar.gzip", "size": 308972, "digest": "sha256:1b23fa3ccba56eced7bd53d424b29fd05cd66ca0203d90165e988fdd8e71fed7" }, { "mediaType": "application/vnd.docker.image.rootfs.diff.tar.gzip", "size": 17711780, "digest": "sha256:b714494d7662fbb89174690cefce1051117ed524ec7995477b222b8d96fb8f0c" }, { "mediaType": "application/vnd.docker.image.rootfs.diff.tar.gzip", "size": 1776866, "digest": "sha256:1098418f3d2f83bcccdbb3af549d75d9a5e9c37420e5c0d474fd84b022f6c995" }, { "mediaType": "application/vnd.docker.image.rootfs.diff.tar.gzip", "size": 360, "digest": "sha256:ca727cee7c2469ab6edb7ca86378985b3747938a325ddb7d90f3b85a3d14b34f" }, { "mediaType": "application/vnd.docker.image.rootfs.diff.tar.gzip", "size": 4315553, "digest": "sha256:767bbe8ba063767093b0350e8d1b86b438c677e3884c2a21851c00970d88317c" } ] } }, { "Ref": "docker.io/onap/app-py-arm-linux:latest", "Descriptor": { "mediaType": "application/vnd.docker.distribution.manifest.v2+json", "digest": "sha256:8a06f997353177dae82d7e01bc3893b2f05c6ac27b317655df3ca2287f9b83a9", "size": 1786, "platform": { "architecture": "arm64", "os": "linux" } }, "SchemaV2Manifest": { "schemaVersion": 2, "mediaType": "application/vnd.docker.distribution.manifest.v2+json", "config": { "mediaType": "application/vnd.docker.container.image.v1+json", "size": 6864, "digest": "sha256:b097a21c92a9a0dde06f9b36bf10def56a028b3b9b1617d6d6c6a8559d14d9d7" }, "layers": [ { "mediaType": "application/vnd.docker.image.rootfs.diff.tar.gzip", "size": 2099514, "digest": "sha256:47e04371c99027fae42871b720fdc6cdddcb65062bfa05f0c3bb0a594cb5bbbd" }, { "mediaType": "application/vnd.docker.image.rootfs.diff.tar.gzip", "size": 176, "digest": "sha256:b4103359e1ecd9a7253d8b8a041d4e81db1ff4a5e1950bc0e02305d221c9e6c2" }, { "mediaType": "application/vnd.docker.image.rootfs.diff.tar.gzip", "size": 308518, "digest": "sha256:92079a442932f09a622f4a0f863e5cc2f6e0471a98e5121fa719d2a276440386" }, { "mediaType": "application/vnd.docker.image.rootfs.diff.tar.gzip", "size": 18605961, "digest": "sha256:f1fc35806f46347993a2cd1eb7f7dd7837b0bef0392c8e2c973b24c02ad874a9" }, { "mediaType": "application/vnd.docker.image.rootfs.diff.tar.gzip", "size": 1786389, "digest": "sha256:c2983ee3d71107a9a0bc1996fc3a58e050026995fad4aee9d72539153db1df3d" }, { "mediaType": "application/vnd.docker.image.rootfs.diff.tar.gzip", "size": 391, "digest": "sha256:44c3eae5ed66bb040727a64fd78573fe6cc4a94a9317d5cd6f39e53332c2ae21" }, { "mediaType": "application/vnd.docker.image.rootfs.diff.tar.gzip", "size": 4305558, "digest": "sha256:8daa79c3024a565c320ff69990ad48273937cc3f6f0cdb324e086c268cf6245e" } ] } } ]
Building Multi-CPU container images using CI/CD pipelines
The following diagram depicts a CI/CD flow that implements the production of multi-cpu architecture container images.
Although the flow shows a pipeline for two branches: arm-linux and intel-linux, the model can be extended --given the hardware and software resources-- to any number of platforms.