January 2022 Summaries

Prometheus has gained significant popularity as a mature CNCF project, with widespread adoption by over 500,000 companies and a substantial presence on GitHub. To ensure compatibility, especially with long-term storage and third-party services, Prometheus co-creator Julius Volz initiated compatibility testing against PromQL and Prometheus itself, which expanded to include OpenMetrics. This led to the development of the Prometheus Conformance program in collaboration with Richard Hartmann, aiming to standardize compatibility testing for Prometheus, PromQL, and Remote Write. The initiative ensures vendors maintain compliance with evolving Prometheus standards, as demonstrated by Logz.io's implementation, which shows 100% PromQL compatibility, supporting the reuse of existing dashboards and scripts. This effort underscores the commitment to maintaining observability at scale, supported by open-source collaboration, while guarding against false compatibility claims by vendors.

Windows applications play a crucial role in many organizations, necessitating support for Windows containers when transitioning to Kubernetes-based architectures. Historically, the lack of Windows container support limited options to Linux container images until Windows containers became available, enabling more seamless DevOps practices for Windows-based applications. Observability of Windows Kubernetes clusters is complex due to the absence of privileged permissions required for metrics collection. Current solutions, like those by aidaspsibr and Octopus, involve installing extensions and using reverse proxies, but they lack generic applicability across all clusters. A novel solution utilizing OpenTelemetry and a Helm chart has been proposed to facilitate metrics collection on Windows nodes, involving a privileged Linux container for SSH connections, a reverse proxy for endpoint exposure, and an OpenTelemetry collector for metrics forwarding. While effective, this approach requires careful handling of credentials and periodic job execution to accommodate new nodes, with options like CronJobs for scheduling, and presents challenges related to job automation and pod management. Despite these challenges, this solution offers a customizable framework for integrating Windows containers into Kubernetes monitoring, with open-source resources available to adapt the process for different backends or specific user needs.