etcd

Start a single-node etcd cluster

$ etcd

$ etcd --advertise-client-urls [http://127.0.0.1:1234] --listen-client-urls [http://127.0.0.1:1234]

$ etcd --name [my_etcd_cluster]

$ etcd --enable-pprof --metrics extensive

etcd [--config-file path] [flags]

--name string
    Human-readable name for this member (default: default)

--data-dir string
    Directory to store data (default: default.etcd)

--listen-client-urls urls
    URLs to listen for client traffic (default: http://localhost:2379)

--advertise-client-urls urls
    Client URLs to advertise to cluster (default: http://localhost:2379)

--listen-peer-urls urls
    URLs to listen for peer traffic (default: http://localhost:2380)

--initial-advertise-peer-urls urls
    Peer URLs to advertise (default: http://localhost:2380)

--initial-cluster cluster-config
    Initial cluster config (e.g., default=http://localhost:2380)

--initial-cluster-state new|existing
    Bootstrap state (default: new)

--initial-cluster-token string
    Cluster token for bootstrap (default: etcd-cluster)

--heartbeat-interval duration
    Raft heartbeat interval (default: 100ms)

--election-timeout duration
    Raft election timeout (default: 1s)

--snapshot-count uint
    Wal entries per snapshot (default: 10000)

--client-cert-auth
    Enable client cert auth

--cert-file path
    TLS cert path

--key-file path
    TLS key path

--trusted-ca-file path
    CA cert path

--peer-cert-file path
    Peer TLS cert

--peer-key-file path
    Peer TLS key

--peer-trusted-ca-file path
    Peer CA cert

--auto-compaction-mode mode
    Compaction mode: periodic|revision|background

--auto-compaction-retention duration
    Retention period for compaction

--max-request-bytes uint
    Max request size (default: 1.5MB)

--experimental-enable-v2
    Enable v2 API (deprecated)

--config-file path
    Path to config YAML/JSON

--force-new-cluster
    Force new cluster from existing data

--enable-v2
    Enable v2 support (legacy)

--version
    Print version

--help
    Show help

etcd is a distributed, reliable key-value store designed for storing critical configuration data in distributed systems. It uses the Raft consensus algorithm to ensure data consistency and high availability across a cluster of nodes. etcd provides a simple HTTP API for watch, put, get, delete, lease, and auth operations, making it ideal for service discovery, configuration management, and coordination.

Primarily used by Kubernetes as the backing store for cluster state (e.g., pods, services, configmaps), etcd is deployed as a cluster (typically 3-5 nodes for production) to tolerate failures. It supports features like TLS encryption, authentication (RBAC), snapshots for backups, compaction for storage efficiency, and alarms for resource limits. etcd runs as a standalone binary on Linux, listening on client (default: 2379) and peer (2380) ports. It's written in Go, lightweight, and supports proxy and gateway modes for scalability.

For production, configure via flags, YAML files, or systemd units. Always bootstrap with proper initial cluster config to avoid split-brain issues.

Requires cluster bootstrap; single node unsuitable for production. High disk I/O on writes; monitor alarms. TLS recommended. Avoid running as root.

Use --initial-cluster with all members for new clusters. Match advertise URLs across nodes.

Use etcdctl snapshot save for consistent backups; restore with etcdutl snapshot restore.

Default store for API server. Tune for large clusters: increase --quota-backend-bytes.

Developed by CoreOS in 2013 as Raft-based store. v3.0 (2016) introduced gRPC API, backward-incompatible with v2. Donated to CNCF 2018. Now at v3.5+ with gRPC gateway, better compaction.

etcdctl(1), systemd(1), kubectl(1)