May 2026 Summaries

Zerops Cloud Platform (ZCP) is designed to provide AI coding agents with a robust development environment that mirrors the real production infrastructure typically used by senior developers, rather than relying on sandboxes or mock environments. By inverting the traditional approach of integrating infrastructure onto the agent, Zerops prioritizes building a solid cloud platform foundation first, utilizing its own orchestration and system containers to ensure scalability and parity across development, staging, and production environments. This setup allows AI agents to interact with the same managed services and deployment pipelines that human developers would use, eliminating the "it works on my machine" problem and encouraging seamless collaboration between human and AI developers. Users have the flexibility to bring their own agents and subscriptions, while Zerops provides the infrastructure without vendor lock-in, supporting various frameworks and enhancing the development process through its Cloud IDE and other customizable tools. With a focus on enhancing rather than replacing human developers, Zerops empowers both agents and developers to work more effectively together on real production infrastructure.

Zerops has implemented a dedicated logging stack for each project on its platform to ensure efficient, scalable, and reliable logging infrastructure, initially experimenting with InfluxDB but finding it unsuitable due to its high resource demands. The company then developed a custom Go service using SQLite, which worked for a while but faced limitations in storage and query performance. Seeking a more robust solution, Zerops adopted VictoriaLogs, which offered vastly improved storage capabilities, query speed, and compression efficiency. This transition was complemented by the integration of Vector for log ingestion and transformation, syslog-ng for log forwarding, and zlogproxy to maintain compatibility with existing tools. The migration to VictoriaLogs was executed smoothly and without downtime across thousands of projects, significantly enhancing in-project logging capabilities. However, Zerops advises users to forward logs to external services for critical production observability, ensuring logs remain accessible even during infrastructure failures.

CVE-2026-31431, known as "Copy Fail," is a critical vulnerability affecting most Linux systems since 2017, allowing an unprivileged process to execute a controlled 4-byte write into the kernel's page cache, leading to potential root access. The vulnerability is particularly worrisome for shared-host environments due to its deterministic, cross-tenant, and invisible nature. Zerops, using shared-kernel Incus system containers for performance and cost efficiency, quickly addressed the issue by leveraging its infrastructure design that tracks the Zabbly mainline kernel, allowing them to deploy a patched kernel across their nodes within hours of the vulnerability's disclosure. This swift response contrasted with the challenges faced by individual server managers and platforms relying on older kernels, which required coordination for patches and maintenance. The incident underscores the importance of high availability (HA) services in maintaining uptime during such kernel emergencies, with Zerops HA services experiencing zero disruption while non-HA services had minimal downtime.