This article covers single-node validation and three-node production deployment with cephadm. For architecture, operations, and functional testing, see Ceph Architecture and Operations. 1. Disk Preparation An OSD can only use a block device (/dev/sdX, a partition, /dev/loopN). 1.1 Inspecting Available Devices 1 2 3 4 5 6 7 lsblk -o

Ceph is a unified distributed storage system that supports block storage (RBD), object storage (RGW), and file storage (CephFS). The official cephadm can orchestrate each daemon as a container (Docker, or Podman + Containerd); see Deploying Ceph in Containers for details. 1. Architecture 1.1 Topology and Data Paths The diagram below shows a common five-node topology (3 MON + 2 MGR + 5 OSD): MONs only need to come in odd numbers, and in a five-node deployment 3 is enough (tolerating one MON failure); 5 MONs tolerate 2 failures, but the metadata overhead is larger, so there is generally no need to run a MON on every machine.

VictoriaMetrics (VM for short) is a time-series database compatible with the Prometheus ecosystem. It uses fewer resources than Prometheus and comes in both single-node and cluster editions. The single-node edition can directly replace Prometheus storage; the cluster edition splits reads, writes, and storage across vmstorage / vminsert / vmselect, giving it stronger horizontal scaling.

1. What Agent Skills Is Agent Skills is a production-grade engineering skill pack for AI coding assistants. It encodes the way senior engineers work — their working methods, quality gates, and best practices across the full Define → Plan → Build → Verify → Review → Ship lifecycle — into Markdown workflows (Skills) that an agent can execute.

1. What Headroom Is Headroom is a context compression layer for AI agents. Before the LLM receives a request, it performs content-aware compression on tool output, logs, RAG chunks, file contents, and conversation history, cutting token usage to 5%–40% of the original while keeping answer quality as unchanged as possible.