1 views collected around this technical thread.
This article describes how the OPS team designed and implemented a private‑cloud Elasticsearch service using Mesos for resource management, Docker containers orchestrated by Marathon, and a suite of monitoring, self‑service configuration, and continuous deployment tools to improve resource utilization and operational efficiency.
This article describes how a team migrated Spark 1.6.x running on Mesos to a Marathon‑Docker based architecture that provides dynamic executor scaling, resolves configuration and resource‑allocation issues, and improves monitoring, fault‑tolerance, and upgrade processes for large‑scale streaming workloads.
This article explains the Mesos distributed system kernel, its resource‑allocation workflow, and how Qunar engineers applied and evolved Mesos, Marathon, and custom frameworks to achieve fine‑grained scheduling, high availability, service discovery, and multi‑tenant management in a large‑scale production environment.
This article walks through installing a Mesos‑based distributed scheduling platform on macOS using Docker containers, covering custom image creation, registry configuration, Docker‑VM quirks, environment‑variable setup for Zookeeper, Mesos master and slaves, Marathon deployment, and practical tips for accessing containers via socat, Portainer, chroot, and nsenter.
This article describes how Qunar's operations team designed and implemented a cloud‑native Elasticsearch‑as‑a‑Service platform using Mesos, Marathon, and Docker, covering requirements analysis, technology selection, resource quota management, cluster isolation, service discovery, data reliability, monitoring, automated deployment, and future improvements.
This article details how Ping An Technology designed and operated the Padis platform—an internal Docker‑based PaaS that evolved to Mesos + Marathon, providing independent container IPs, automated networking, load balancing, CMDB integration, monitoring, and log aggregation to support thousands of containers in production financial applications.
This article explains the Mesos distributed system kernel, its master‑slave architecture, fine‑grained resource scheduling, and how Qunar leverages Mesos and Marathon for log processing, Spark, Alluxio, and multi‑tenant services while addressing framework unification, HA, service discovery, and operational challenges.
This article describes how Qunar's DevOps team built a Mesos‑powered log analysis platform using ELK, Marathon, Chronos, and custom monitoring frameworks, detailing deployment architecture, resource management, container logging, image handling, and operational lessons learned.