Install Orchestrators

Orchestrators (Pro)

Requires: replication-manager-pro flavor

The pro flavor of replication-manager can provision and manage the full lifecycle of database clusters — bootstrapping containers, deploying configuration, starting/stopping services — by delegating to an external orchestrator.

Set the active orchestrator with:

prov-orchestrator = "opensvc"   # opensvc | kube | slapos | onpremise | local
Orchestrator prov-orchestrator value Description
OpenSVC opensvc Service orchestration cluster — the primary supported orchestrator. Default for pro.
Kubernetes kube Container orchestration via a K8s cluster.
SlapOS slapos Distributed service mesh from Nexedi.
On-Premise SSH onpremise SSH-based bootstrap onto existing bare-metal or VM hosts. No orchestrator agent required.
Local local Bootstrap processes directly on the same host as replication-manager. For testing only.

osc supports onpremise and local only. All five modes are available in pro.

OpenSVC

OpenSVC is a service orchestration cluster that manages Docker/Podman micro-services across a pool of agents. replication-manager uses the OpenSVC API to push service definitions and trigger provisioning.

Install the OpenSVC Agent

On each database host, install the OpenSVC agent:

# Debian/Ubuntu
curl -o /tmp/opensvc.latest.deb https://repo.opensvc.com/deb/current
dpkg -i /tmp/opensvc.latest.deb

# RHEL/CentOS
curl -o /tmp/opensvc.latest.rpm https://repo.opensvc.com/rpm/current
rpm -ivh /tmp/opensvc.latest.rpm

Start and join the cluster:

om daemon start
om cluster set --param cluster.secret --value <shared-secret>
om daemon join --node <first-node-ip>

Full OpenSVC documentation: docs.opensvc.com

Configure replication-manager for OpenSVC

replication-manager can use either the Cluster API (recommended, direct agent HTTP2 API) or the legacy Collector API (Signal18 SaaS or on-premise collector).

Cluster API (recommended)

Point replication-manager at any OpenSVC cluster node:

prov-orchestrator           = "opensvc"
opensvc-cluster-host        = "opensvc-node1:1215"
opensvc-cluster-admin-user  = "[email protected]:opensvc"

Collector API (legacy)

prov-orchestrator    = "opensvc"
opensvc-host         = "collector.mycompany.com:9443"
opensvc-admin-user   = "[email protected]:secret"

The Signal18 SaaS collector (ci.signal18.io:9443) can be used for evaluation — contact Signal18 to obtain your account.yaml and place it in the share/opensvc/ directory of replication-manager-pro.

OpenSVC Video Walkthrough

Kubernetes

replication-manager can deploy database services into a Kubernetes cluster using a standard kubeconfig file.

Prerequisites

  • A running Kubernetes cluster (1.20+)
  • kubectl configured and working on the replication-manager host
  • Sufficient RBAC permissions to create namespaces, deployments, services, and PersistentVolumeClaims

Configure replication-manager for Kubernetes

prov-orchestrator = "kube"
kube-config       = "/etc/replication-manager/kube/config"

Copy your cluster's kubeconfig to the path referenced above. replication-manager uses the default context from the file.

cp ~/.kube/config /etc/replication-manager/kube/config
chmod 600 /etc/replication-manager/kube/config

SlapOS

SlapOS is a distributed service mesh used in the Nexedi ecosystem. replication-manager-pro can deploy database instances as SlapOS services.

Configure replication-manager for SlapOS

prov-orchestrator = "slapos"

Contact Signal18 for SlapOS deployment templates specific to your environment.

On-Premise SSH

The onpremise mode bootstraps database configuration and services on existing hosts via SSH — no orchestrator agent required. replication-manager connects as a sudoer user and deploys configuration files, starts/stops services using the host's init system.

Prerequisites

  • SSH access from the replication-manager host to each database host
  • A user with passwordless sudo on the database hosts
  • MariaDB or MySQL server packages already installed on the database hosts

Configure SSH access

prov-orchestrator         = "onpremise"
onpremise-ssh             = true
onpremise-ssh-port        = 22
onpremise-ssh-credential  = "deploy"

replication-manager uses its own SSH private key (~/.ssh/id_rsa by default). Add the corresponding public key to the authorized_keys of the deploy user on each database host.

Local (Testing Only)

The local orchestrator boots MariaDB or MySQL processes directly on the same host as replication-manager. It is intended for local regression testing and benchmarking — not for production use.

Prerequisites

  • MariaDB or MySQL server binaries installed locally
  • HAProxy and/or ProxySQL if proxy testing is needed
  • Sysbench for load injection
prov-orchestrator       = "local"
prov-db-binary-basedir  = "/usr/bin"
haproxy-binary-path     = "/usr/sbin/haproxy"
proxysql-binary-path    = "/usr/bin/proxysql"

Next Steps

After installing and configuring the orchestrator, proceed to Provisioning Configuration to define service templates and bootstrap your first managed cluster.