etcd

distributed key-value store with Raft consensus

TLDR

Start etcd server

$ etcd

Start with custom data directory

$ etcd --data-dir [/var/lib/etcd]

Start with specific listen address

$ etcd --listen-client-urls [http://localhost:2379]

Start with advertised URLs

$ etcd --advertise-client-urls [http://localhost:2379]

Start a cluster member

$ etcd --name [node1] --initial-cluster [node1=http://host1:2380,node2=http://host2:2380]

Enable TLS

$ etcd --cert-file [cert.pem] --key-file [key.pem]

etcd is a distributed key-value store that provides reliable, consistent data storage for distributed systems. It implements the Raft consensus algorithm to ensure strong consistency across multiple nodes, making it suitable for critical configuration data and service coordination.The server forms the backbone of Kubernetes cluster state management and is used extensively in cloud-native architectures for configuration management, service discovery, and distributed locking. Its simple HTTP/gRPC API and watch functionality enable applications to respond to configuration changes in real-time.etcd prioritizes consistency and availability, making it ideal for storing cluster membership, feature flags, and other distributed system metadata.

PARAMETERS

--name name

Human-readable node name.

--data-dir path

Data directory path.

--listen-client-urls urls

Client listen URLs.

--advertise-client-urls urls

Advertised client URLs.

--listen-peer-urls urls

Peer listen URLs.

--initial-cluster config

Initial cluster configuration.

--initial-cluster-state state

Initial cluster state (new or existing).

--cert-file file

TLS certificate file.

--key-file file

TLS key file.

--initial-cluster-token token

Unique cluster token to prevent cross-cluster interaction.

--initial-advertise-peer-urls urls

Peer URLs to advertise to the rest of the cluster.

--snapshot-count count

Number of committed transactions to trigger a snapshot to disk (default 100000).

--quota-backend-bytes bytes

Raise alarms when backend size exceeds the given quota (0 defaults to low space quota).

--max-request-bytes bytes

Maximum client request size in bytes the server will accept (default 1572864).

CONFIGURATION

/etc/etcd/etcd.conf.yml

Main configuration file for etcd server settings, cluster topology, and security options. Overrides all command-line flags and environment variables when specified.

ETCD_*

Every flag has a corresponding environment variable prefixed with ETCD in all caps and snake case (e.g., --data-dir becomes ETCDDATA_DIR). Command-line flags take precedence over environment variables.

CAVEATS

Requires careful cluster planning for production use. Network partitions affect availability. Disk performance impacts latency. Regular backups essential. Raft consensus requires quorum majority.

HISTORY

etcd was created by CoreOS in 2013 and became a Cloud Native Computing Foundation (CNCF) project. It has become the de facto standard for Kubernetes configuration storage and is widely deployed in production cloud environments.