January 2018 Summaries

Between December 12 and December 17, Semaphore experienced network instabilities due to a faulty device at a Tier 1 network provider, affecting communication with platforms like GitHub and Docker. The issue began with high network latency and sporadic TLS handshake timeouts, prompting Semaphore to collaborate with its hosting provider to trace the root cause, which was initially elusive. Despite efforts such as altering routing paths, deploying proxy servers, and running comprehensive network tests, the problem persisted until collaboration with GitHub led to the detection and replacement of a faulty line card by Telia, a network operator. Semaphore's response included setting up a new build cluster and planning for future on-demand multi-region provisioning to mitigate similar issues swiftly. The company acknowledged the importance of reliable service and committed to enhancing monitoring and rapid response capabilities to prevent future disruptions.

In an interview with Alban Crequy, co-founder and CTO at Kinvolk GmbH, insights into the rkt container engine reveal its design as a secure, pod-native container solution for Linux, emphasizing composability and alignment with existing technologies like systemd and TPM. Crequy discusses the project's success in community contributions, acknowledging challenges in managing fluctuating contributor numbers and open issues, a common issue in open-source projects. He compares rkt to Docker, highlighting rkt's flexibility in swapping isolation environments and its daemonless architecture, which appeals to systems architects. The conversation also touches on the transition from ACI to OCI image standards, the influence of CoreOS's Container Linux philosophy on other operating systems, and the use of Semaphore for testing due to its support for a full VM environment. Crequy expresses enthusiasm for technologies like eBPF and Kubernetes, indicating a bright future for containerization and the potential for innovative developments in the space.

Emile Vauge, the creator of Traefik, an open-source reverse-proxy written in Go, discusses the project's integration with modern software infrastructures such as Kubernetes, Amazon ECS, and Docker Swarm. Traefik is designed to manage ingress traffic dynamically, and its maintainers, split between Containous and external contributors, promote an inclusive environment for contributions by simplifying the maintenance process and automating tasks with bots. Issues and feature requests are managed efficiently through GitHub tools, and the project’s versions are whimsically named after French cheese, reflecting its origins. Despite challenges like managing pull request merges, a bot now automates the process, allowing the team to maintain a release cycle of every two months. Semaphore has significantly improved their CI times, reducing test durations from over 50 minutes to under 10, enabling quicker feedback on pull requests. Vauge also emphasizes the importance of being involved and responsive when open-sourcing company projects and expresses enthusiasm for improvements in Go dependency management with projects like dep.