Cluster Architecture & Operations

Lessons in this group, roughly in build order:

• node-roles-master-data-ingest-coordinating — Every Elasticsearch node advertises a set of roles in node.roles; the roles a node holds determine which…
• discovery-cluster-formation — Discovery is how nodes find each other and elect a master that owns the authoritative cluster state;…
• shard-allocation-routing — Allocation is the master’s decision of which node hosts each shard copy; routing is the deterministic…
• cluster-health-green-yellow-red — Cluster health is a single traffic-light status — green, yellow, or red — computed from whether every…
• index-lifecycle-management-ilm — ILM is a policy engine that automatically moves an index through phases — hot, warm, cold, frozen, delete…
• snapshots-restore — A snapshot is an incremental backup of indices (and optionally cluster state) written to a repository on…
• rolling-upgrades — A rolling upgrade moves a cluster to a new Elasticsearch version one node at a time, with no downtime, by…