Feature Request
The Docker image (ghcr.io/miantiao-me/broadcastchannel:main) is currently only built for linux/amd64. On ARM64 hosts, Docker falls back to QEMU userspace emulation at runtime, which causes noticeable CPU overhead and degraded performance.
Background
I'm running BroadcastChannel via Docker on an ARM64 server. Because the image only ships amd64, Docker has to use QEMU to translate every instruction at runtime, which increases system load unnecessarily.
ARM64 servers are becoming more common across cloud providers (AWS Graviton, Oracle Cloud ARM free tier, Ampere Altra) and for self-hosting (Apple Silicon, Raspberry Pi). Supporting arm64 natively would benefit anyone deploying on these platforms.
Proposed Solution
The existing Dockerfile uses node:lts-alpine, which already provides multi-architecture base images — so no Dockerfile changes are needed. Only 3 lines need to be added to .github/workflows/docker.yml to set up QEMU + Docker Buildx and specify platforms: linux/amd64,linux/arm64.
I've submitted a PR with the implementation: #135
Feature Request
The Docker image (
ghcr.io/miantiao-me/broadcastchannel:main) is currently only built forlinux/amd64. On ARM64 hosts, Docker falls back to QEMU userspace emulation at runtime, which causes noticeable CPU overhead and degraded performance.Background
I'm running BroadcastChannel via Docker on an ARM64 server. Because the image only ships
amd64, Docker has to use QEMU to translate every instruction at runtime, which increases system load unnecessarily.ARM64 servers are becoming more common across cloud providers (AWS Graviton, Oracle Cloud ARM free tier, Ampere Altra) and for self-hosting (Apple Silicon, Raspberry Pi). Supporting
arm64natively would benefit anyone deploying on these platforms.Proposed Solution
The existing
Dockerfileusesnode:lts-alpine, which already provides multi-architecture base images — so no Dockerfile changes are needed. Only 3 lines need to be added to.github/workflows/docker.ymlto set up QEMU + Docker Buildx and specifyplatforms: linux/amd64,linux/arm64.I've submitted a PR with the implementation: #135