Project Information
The OpenTelemetry Demo is a sample project that demonstrates a production deployment of OpenTelemetry, a framework for generating telemetry data to be used by observability systems. It consists of over a dozen microservices built in each language supported by the project, and is distributed through Docker containers for deployment locally (via Docker Compose) or to Kubernetes (through Helm). We currently build our images in GitHub Actions, but due to the size of several dependencies (including gRPC), these builds can take 20+ minutes. We would like to offer container images native to different architectures, but building in emulation is prohibitively slow (4+ hour builds).
What We Need
We need the ability to run a Kubernetes cluster with on-demand nodes in order to dynamically create build runners with the appropriate architecture for pull requests and releases of the demo. I believe this would be a dedicated small x86 machine to act as the control plane, and on-demand large x86 and arm nodes (or perhaps a single large x86 node to act as the control plane + worker, with an on-demand arm node for release builds only?)
The OpenTelemetry Demo (https://github.com/open-telemetry/opentelemetry-demo) repository needs x86 and ARM build runners in order to build native container images.
Name: Austin Parker E-Mail: austin@ap2.io Organization: OpenTelemetry
Project Information The OpenTelemetry Demo is a sample project that demonstrates a production deployment of OpenTelemetry, a framework for generating telemetry data to be used by observability systems. It consists of over a dozen microservices built in each language supported by the project, and is distributed through Docker containers for deployment locally (via Docker Compose) or to Kubernetes (through Helm). We currently build our images in GitHub Actions, but due to the size of several dependencies (including gRPC), these builds can take 20+ minutes. We would like to offer container images native to different architectures, but building in emulation is prohibitively slow (4+ hour builds).
What We Need We need the ability to run a Kubernetes cluster with on-demand nodes in order to dynamically create build runners with the appropriate architecture for pull requests and releases of the demo. I believe this would be a dedicated small x86 machine to act as the control plane, and on-demand large x86 and arm nodes (or perhaps a single large x86 node to act as the control plane + worker, with an on-demand arm node for release builds only?)