Everything ElastiCluster does, and where each part stands
The complete list of capabilities, grouped by area. Each one is marked as working in development builds, being
built now, or planned for a named release. Nothing has been released yet.
- Working in development builds
- Implemented and tested in the current development code.
- In development now
- Being implemented in the current milestone.
- Planned for a named release
- Designed and scheduled for the named release.
Servers and agents
More about servers and agents- Add servers over SSH Host-key confirmation; the adding admin approves the server. Working in development builds
- One-line install with join tokens Single-use or fleet tokens, bound to network, hostname or machine ID; admin approval required. Working in development builds
- Preflight checks and one-click fixes About twenty checks, blocking or warning, with remediations that respect the host’s local policy. Working in development builds
- Agent in a container or as a binary Dials out over mutual TLS; signed self-updates without restarting databases. Working in development builds
- Data identity guard A database refuses to start on someone else’s data or an unmounted disk. Working in development builds
- Drain, re-address Move work off a server or change its address through a plan. Planned for v0.1
- OS patching in waves Updates and reboots that never take two members of one cluster at once. Planned for v0.8
- Proxies and offline bundles HTTP proxies for agents and air-gapped installs from signed bundles. Planned for v0.8
Plans, approvals and operations
More about plans, approvals and operations- Plans with diff, risk and disruption Every change previewed before it runs, the same in console, CLI, API and MCP. In development now
- Policies and approvals Per-environment rules, four-eyes, approver roles, typed confirmation for critical plans. In development now
- Stale approvals and re-plan Approvals stop being valid when the cluster or policy changes. In development now
- Maintenance windows Disruptive steps wait for the window, with time zones and daylight-saving rules. In development now
- Resumable operations Retry, skip or abort a failed step; resume after a control-plane restart. In development now
- Templates Versioned starting points for new clusters; policy always wins over template values. In development now
High availability and endpoints
More about high availability and endpoints- PostgreSQL with Patroni and etcd Three members or two plus a witness; synchronous replication presets. Planned for v0.1
- HAProxy endpoints and virtual IP rw, ro and rw-direct for PostgreSQL, with PgBouncer pooling and keepalived. Planned for v0.1
- Write canary and health conditions A real write through the endpoint every 30 seconds; read, write and quorum tracked apart. Planned for v0.1
- MongoDB, Valkey and Dragonfly Replica-set elections, Sentinel, and Dragonfly replication in beta. Planned for v0.4
- MariaDB Galera Quorum-based clusters with a preferred-writer endpoint. Planned for v0.5
- Dedicated proxy servers Shared HAProxy servers with one virtual IP pair for several clusters. Planned for v0.7
Backups and recovery
More about backups and recovery- Encrypted backups on by default pgBackRest for PostgreSQL to S3, filesystem or SFTP repositories. Planned for v0.2
- Point-in-time recovery Restore to a moment, a transaction or an automatic restore point, into a new cluster. Planned for v0.2
- Restore drills Scheduled test restores that measure the restore time. Planned for v0.2
- Forks Short-lived copies of a cluster that delete themselves after seven days. Planned for v0.2
- Signed manifests and least privilege Servers cannot delete history; tampered backups are rejected. Planned for v0.2
- Second copy, DR standby, delayed replicas Immutable second repository and cross-site PostgreSQL standby clusters. Planned for v0.7
Monitoring and alerts
More about monitoring and alerts- Host metrics and log streaming Agents collect host metrics and stream logs. Working in development builds
- Built-in metrics with PromQL An embedded time-series database, per-engine charts and Grafana dashboards. Planned for v0.3
- Alert rules with runbooks About 100 rules, each with a runbook; overrides per environment and cluster. Planned for v0.3
- Notifications E-mail, Slack, Discord, Telegram, Teams and signed webhooks. Planned for v0.3
- Reports Availability, backups, failovers and governance, as CSV or PDF. Planned for v0.8
DBA tools
More about dba tools- SQL editor and schema browser Read-only by default on a replica; write mode for people by default, unlocked per session in production. Planned for v0.3
- Sessions, locks and query insights Blocking trees, safe cancel and terminate, pg_stat_statements history. Planned for v0.3
- Maintenance jobs and advisor Scheduled REINDEX, pg_repack and partition upkeep; evidence-based recommendations. Planned for v0.7
- Major upgrades Rehearse on a fork, a named commit boundary, lossless rollback before it. Planned for v0.7
- Import and adopt Dump and restore into a new cluster; adopt PostgreSQL by replication. Planned for v0.7
AI and MCP
More about ai and mcp- Personal access tokens with autonomy Scoped, environment-bound tokens that carry an autonomy level. Working in development builds
- Built-in MCP server Tools for inspection, diagnosis and plans; every REST operation reachable. Planned for v0.3
- Limits per environment and kill switches Propose-only in production by default; one switch stops all AI writes. Planned for v0.3
- Connect an assistant Guided setup and ecl mcp setup for Claude Code, Cursor, Codex, VS Code and Claude Desktop. Planned for v0.3
- OAuth for MCP clients Sign-in for hosted connectors such as claude.ai. Planned for v1.1
Security and access
More about security and access- Setup token and two-factor sign-in No default credentials; argon2id passwords, authenticator codes, recovery codes. Working in development builds
- Roles, sudo mode and scoped tokens Five roles, re-authentication for sensitive actions, tokens limited by scope and IP. Working in development builds
- Secrets vault and internal CA Per-secret encryption with rotatable master keys; a separate authority for each cluster once clusters arrive. Working in development builds
- Tamper-evident audit log Hash-chained entries, signed anchors, export to JSONL, syslog, webhook or S3. Working in development builds
- Invitations and break-glass access Resets that never bypass two-factor; emergency access that is always audited. In development now
- Single sign-on and passkeys OpenID Connect providers, passkeys, teams and service accounts. Planned for v0.8
Platform and governance
More about platform and governance- Projects and environments Policies, quotas and an egress allowlist per project and environment. In development now
- REST API OpenAPI 3.1, stable error codes and live updates over server-sent events. Working in development builds
- ecl command line Contexts, browser sign-in, apply and diff, plans and operations from the terminal. In development now
- Schedule calendar Backups, drills, windows, patching and certificate renewals with conflicts shown. Planned for v0.8
- Webhooks Signed CloudEvents with retries and replay. Planned for v0.8
- One-line installer A single command to install the control plane, online or offline. Planned for v0.8
- Terraform provider Clusters as code with Terraform or OpenTofu. Planned after v1.0
Engine-specific details are on the engine pages, and the order of releases is on the roadmap.
Try the console before it ships
The interactive demo runs the whole console on simulated data. Trigger a failover, approve an AI assistant’s plan, restore a backup to a point in time. Nothing to install.