DATABASE FLEET

Provision the database, then keep it alive

Database Fleet is the part of ClusterEye that is not monitoring: it provisions PostgreSQL, MongoDB and MySQL clusters onto your own servers, and then operates them — adding members, rolling upgrades, switchovers, credential rotation and backups. By the time provisioning finishes the cluster is already monitored: installing the ClusterEye agent is the last step of the deploy, so there is no gap between provisioning and monitoring for someone to wire up by hand.

What it provisions

Production topologies are available today for PostgreSQL, MongoDB and MySQL.

PostgreSQL

AVAILABLE
·Patroni-managed HA cluster (Patroni + etcd on the cluster hosts)
·Streaming replication (primary plus standbys)
·Standalone
·Installed from the PGDG repositories, with conservative / balanced / high_performance profiles

MongoDB

AVAILABLE
·Replica set
·Sharded cluster (config servers, shard servers, mongos routers)
·Keyfile-based intra-cluster authentication

MySQL

AVAILABLE
·InnoDB Cluster (Group Replication, 3–9 members)
·MySQL Router with read/write split (optional keepalived VIP)
·MySQL 8.4 LTS and 8.0, with conservative / balanced / high_performance profiles

ClickHouse

ON THE ROADMAP
·Will be added on the same step engine as PostgreSQL and MongoDB

Provisioning is the start — the work comes after

Standing a cluster up is the easy part. Database Fleet runs the work its whole life needs through the same step engine and the same audit trail.

Add and remove membersA node joins or safely leaves a replica set or a streaming replication chain, with topology and configuration rewritten to match.
Add and remove shardsSharded MongoDB clusters grow and shrink by whole shards, with config server records and routing updated accordingly.
Rolling version upgradesMembers are upgraded one at a time with health validated after each, so an upgrade never takes the whole cluster down at once.
SwitchoverA planned primary change — a Patroni switchover, a manual promote on streaming replication, or an InnoDB Cluster primary change — runs and is recorded as a single operation.
Credential rotationCluster passwords and internal credentials are rotated, with the dependent configuration updated in the same pass.
BackupspgBackRest is configured for PostgreSQL, on-demand backups are taken, and the backup policy is changed as an operation of its own.
DecommissioningA deployment that is no longer needed is removed in a controlled way, cleaning up what it left behind.

How it works

Every operation runs through a state machine. When a step fails the job is not left half-done: the completed steps are undone.

01

Preflight

Target servers are checked before anything is deployed. If a blocking issue is found the deployment does not start, and you are told what was found and how to fix it.

02

Deploy

Provisioning executes step by step, with the state and output of each step visible live. No step is a black box.

03

Validation

When provisioning finishes, the cluster is verified to be genuinely healthy — 'the command exited zero' and 'the cluster works' are not the same claim.

04

Rollback on failure

If a step fails, the completed steps are undone automatically and the deployment returns to a consistent state. No half-provisioned server is left behind.

Monitored from the moment it exists

The last step of a deploy installs the ClusterEye agent: a single-use registration token is minted, the agent is installed, and the connection is verified. A new cluster starts sending its first metrics seconds after provisioning completes. There is no separate provisioning tool to wire into a separate monitoring tool — it is the same platform, the same inventory, the same timeline.

Frequently asked questions

PostgreSQL, MongoDB and MySQL today. On PostgreSQL: Patroni-managed HA clusters, streaming replication and standalone. On MongoDB: replica sets and sharded clusters. On MySQL: InnoDB Cluster with MySQL Router. ClickHouse is on the roadmap and is not available yet.

Don't hand-build the next cluster.

Book a demo to see Database Fleet in your own environment.

Book a demoProduct guide