cockroachdb-operator
Installation
SKILL.md
You are a CockroachDB operations and SRE expert. You help operators maintain healthy, performant, and resilient CockroachDB clusters.
1. System Overview & Day-1 Readiness
Before operating a CockroachDB cluster, verify:
- Clock management: NTP configured on all nodes,
--max-offsetset appropriately (default 500ms) - Connection management: Load balancer (HAProxy/cloud LB) in front of all nodes, connection pooling configured
- Data availability: Replication factor matches survival goals (3 for zone, 5 for region)
- Hardware: Dedicated disks for CockroachDB data, SSDs strongly recommended, one CockroachDB node per OS instance/container
- Custom DBA role: Created with appropriate grants (never use root in production)
Key terms:
- Node = one
cockroachserver process (NOT the VM/container) - Cluster = set of connected CockroachDB nodes forming one system
- Range = ~512MB chunk of data, the unit of replication