Arm64 Image

[viseshrp]

Is there one planned?

[krumware]

There is, can you please try calcom/docker:canary and let me know if it works for you?

[viseshrp]

Yes. I see it. Let me try that and report back. Thanks.

[viseshrp]

@krumware Tried this out on a 4gb Raspberry pi 4B and is looking good so far. Thank you. Do let me know if I can help test it out. Could you link me to the source/Dockerfile for this? Thanks!

[krumware]

It's the same dockerfile from this repo, but I needed to set up arm64 build runners for it. Doing the build emulated on x86 or on slower filesystems (like rpi) was a no-go. We'll get this automated, thank you for testing!

[viseshrp]

Excellent. Thanks!

[viseshrp]

@krumware The calcom version in the canary image says 2.7.6 while the current version in Releases here is 2.9.4. Is that intentional? If yes, can you push a new canary with the latest version (or show me how to build one if you can't)? Thanks.

[krumware]

That's correct. I'll update the branch and rebuild it in the meantime

On Fri, Jun 2, 2023, 12:06 PM Visesh Rajendraprasad < @.***> wrote:

@krumware https://github.com/krumware The calcom version in the canary image says 2.7.6 while the current version in Releases here is 2.9.4. Is that intentional? If yes, can you push a new canary with the latest version (or show me how to build one if you can't)?

— Reply to this email directly, view it on GitHub https://github.com/calcom/docker/issues/246#issuecomment-1573973585, or unsubscribe https://github.com/notifications/unsubscribe-auth/AEEARZTET6PC3ITVP6YBG3LXJIFOZANCNFSM6AAAAAAYU26Z6E . You are receiving this because you were mentioned.Message ID: @.***>

[krumware]

Just updating, I'm still working with the build tooling. There are significant performance challenges building the project on emulated architectures (using qemu). It's looking like I'm needing to setup a permanent multi-node buildx setup to build each arch so that the manifest is combined.

In the short term, I think I'll just suffix the build image tags with -arm to make releases available, but that's not the desired long-term pattern.

[viseshrp]

Appreciate it @krumware . Thank you. Fwiw, I'm now building releases locally (after checking out the calcom submodule) on my Pi 4GB and then running it through compose. Works pretty well for me but it is SLOW. Not too bad though

[krumware]

@viseshrp take a peek at the update dockerfile in feature/multi-arch https://github.com/calcom/docker/blob/feature/multi-arch/Dockerfile

The inclusion of --platform=$BUILDPLATFORM might speed that up significantly. (although that's the nature of the beast, this is a BIG app build)

also, good article here: https://www.docker.com/blog/faster-multi-platform-builds-dockerfile-cross-compilation-guide/

The multi-arch build isn't as straight forward in general here due to the depth of tooling and size of the app build, but single arch is straight forward with those modifications

[viseshrp]

@krumware ah, I see. Thanks! I'll check those out.

[Amorfx]

I see.

What image should I use to run the application on my macbook pro ARM ? For the last version.

[Foxdynamics]

Hi there,

are there any updates on this yet?

I tried today to customize the Jitsi_URL to my own JitsiMeet server and it wouldn't work until I saw that I'm only on version 2.7.6 due to the canary build on the Raspberry Pi 4 Model B.

I would really appreciate it if it were possible to create an up-to-date build for arm64 images.

Regardless, many thanks for this great project!

[kaumudpa]

I have created an ARM64 Image and pushed it to Dockerhub - kaumudpa/calcom - Version 4.0.7