GitHub
Getting startedInstall CobaltDB and run your first SQL query in Go.SQL & JSONTables, indexes, parameterized queries, JSON and transactions.ArchitectureUnderstand the query engine, B+ tree storage and write-ahead log.Server & clientsRun the standalone server and connect over the MySQL wire protocol.Production guidePlan TLS, security, backups, observability and deployment boundaries.Performance benchmarksB+ tree and SQL benchmark results, methodology and hardware.FeaturesSQL, transactions, encryption and a pure Go runtime.Brand assetsDownload the CobaltDB SVG logo and icon.
K to open ESC to close
Documentation
CobaltDBDocumentationProduction guide

Production guide

Plan TLS, security, backups, observability and deployment boundaries.

Know the deployment boundary

CobaltDB currently describes itself as a production-oriented, single-node database candidate. Treat readiness as a property of your application, workload, driver and recovery process.

Replication is not automatic HA. The replication transport does not by itself provide automatic failover or high availability. Plan recovery and failover procedures explicitly.

Read the upstream production guide and HA and failover boundaries before choosing a deployment architecture.

Network & data security

  • Bind listeners to trusted interfaces and keep them private by default.
  • Configure TLS for non-loopback wire access and validate client compatibility.
  • Set a strong, unique admin password through your secret-management system.
  • Evaluate AES-256-GCM encryption, RLS policies and audit logging for your data.
  • Manage encryption keys independently from the data they protect.

Non-loopback replication requires authenticated TLS. Mutual TLS and certificate-name overrides are available for relevant topologies. Use the current server configuration as the authority.

Backups & restoration

Use CobaltDB’s supported backup APIs or CLI for online-consistent backups. A filesystem snapshot helper and an online backup are different workflows.

Test restoration against a separate database before relying on a backup. Preserve required keys, metadata and operational configuration as part of your recovery plan.

Refer to the operations runbook for the current procedures and deployment constraints.

Health & observability

The health listener exposes liveness, readiness and metrics endpoints. Keep the operational listener private or behind your own access control.

Local health checksbash
curl http://127.0.0.1:8420/health
curl http://127.0.0.1:8420/ready
curl http://127.0.0.1:8420/metrics/prometheus

Use representative workloads to set latency and resource budgets. Benchmark persistent storage, concurrency, failure recovery and the exact client versions you plan to ship.

Verify the version you deploy

Pin your release, review the changelog, and run the repository’s verification commands in the appropriate development environment.

Repository checksbash
make verify

# Security/concurrency gate; race detector needs a CGO toolchain
make verify-security

These commands are upstream database checks. Website build and browser checks do not certify database production readiness.

Source-aligned guide · CobaltDB v0.6.1Suggest an improvement