etcd: distributed locking and service discovery
summary
etcd claims to give reliable distributed locking and service discovery via an accessible HTTP+JSON API and a Raft-based consensus core. It shows a working discovery service bootstrap for initial leaders, a locksmith semaphore used to throttle cluster reboots, production fixes for log corruption and cloud sync latency, and integrations used by Kubernetes and Fleet. Planned work includes MVCC snapshots, observer read proxies to reduce watch load, and concrete tradeoffs around TTLs and linearizable reads.
AI-written summary · about
chapters (12)
- 0:00Start
- 3:29Data Store API
- 4:22etcd Cluster
- 5:18bootstrapping
- 6:13discovery.etcd.io/6eadeac2
- 7:33bootstrapped
- 23:24Cluster Wide Reboot Lock
- 26:20Cluster Work Scheduling
- 32:30Log files
- 33:58etcd machine naming
- 35:09sync() in the cloud
- 38:01Read-only proxy
