Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[meta] Tracker for container insight arm64 support #114

Closed
6 tasks done
pingleig opened this issue Oct 27, 2020 · 2 comments
Closed
6 tasks done

[meta] Tracker for container insight arm64 support #114

pingleig opened this issue Oct 27, 2020 · 2 comments
Assignees
Labels
arch/arm64 ARM 64 bit area/documentation Improvements or additions to documentation area/k8s Kubernetes aws/eks Amazon Elastic Kubernetes Service component/container-insight component/fluentd os/linux Linux amd64 & arm64 status/under-review Under review

Comments

@pingleig
Copy link
Member

pingleig commented Oct 27, 2020

This is a tracker to direct discussion of existing issues into one place. There is no detailed timeline (yet). Please leave feedback if you are planning to use it or have concern for breaking changes (if any).

Background

EKS Graviton support is GA and cloudwatch-agent already has an arm64 build (not containerized). Fludentd has arm64 dameonset build.

Related Issues

Not related issues

Design

  • Provide a multi arch image on dockerhub like xray-dameon
    • Existing manifest should still work in theory i.e. pull amd64 on amd64 host, haven't tested it with dockerhub though.
  • Provide new repo/tag that contains arm64 amd64 for people using registry/runtime that does not support multi arch image NOTE: Amazon ECR supports it

TODO

  • provide full snippet for workaround
  • official arm64 image
  • Update related manifest in We recommend users to use fluent-bit which has a multi-arch image, for fluentd users need update the image tag manually since the official daemonset image is x86 only.

Workaround

It is possible to build arm image by yourself from official arm64 packages. Running docker buildx or creating an ARM instance on ec2 as a builder box should both work.

You need to modify example manifests to make sure you are

  • Using arm64 version of fluentd image
  • Using node selector arch=arm64 if your eks cluster is mixing amd64 and arm64 node groups. Even if you build a mult-arch image for cloudwatch-agent, fluentd is still specifying arch using image tag.
@pingleig pingleig added area/documentation Improvements or additions to documentation os/linux Linux amd64 & arm64 arch/arm64 ARM 64 bit component/container-insight component/fluentd labels Oct 27, 2020
@pingleig pingleig self-assigned this Oct 27, 2020
@pingleig pingleig added status/proposed Proposed area/k8s Kubernetes aws/eks Amazon Elastic Kubernetes Service labels Oct 27, 2020
pingleig added a commit to pingleig/amazon-cloudwatch-agent that referenced this issue Dec 8, 2020
* For aws#114 user no longer need to write their own Dockerfile and it
works with `docker buildx`
* Add containerized build for developer w/o a local go environment
@pingleig pingleig added status/under-review Under review and removed status/proposed Proposed labels Dec 23, 2020
@pingleig
Copy link
Member Author

pingleig commented Jan 1, 2021

The arm64 image is out on dockerhub and public ECR

@pingleig
Copy link
Member Author

The public doc is updated (somehow it seems it is out of sync w/ the github repo). If you are using the agent only, there is no need to change anything, the docker image is multi arch so it works out of box. If you use fluentd, you need to update the image tag to match the arch See install fluentd. The recommended approach is using fluent-bit.

Related links

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
arch/arm64 ARM 64 bit area/documentation Improvements or additions to documentation area/k8s Kubernetes aws/eks Amazon Elastic Kubernetes Service component/container-insight component/fluentd os/linux Linux amd64 & arm64 status/under-review Under review
Projects
None yet
Development

No branches or pull requests

1 participant