Group Replication

MySQL Group Replication (GR) is MySQL’s built-in multi-master clustering technology. It provides automatic failover, conflict detection, and distributed recovery without external tools. dbdeployer makes it easy to spin up GR clusters for testing and development.

Minimum version: MySQL 5.7.17+

Single-Primary Mode

In single-primary mode, one node is the primary (read/write) and the rest are secondaries (read-only). Failover is automatic — if the primary fails, the group elects a new one.

dbdeployer deploy replication 8.4.8 --topology=group --single-primary

This creates three nodes by default:

~/sandboxes/group_sp_msb_8_4_8/
├── node1/    # primary
├── node2/    # secondary
├── node3/    # secondary
├── check_nodes
├── start_all
├── stop_all
└── use_all

Connect to the primary:

~/sandboxes/group_sp_msb_8_4_8/n1 -e "SELECT @@port, @@read_only"

Multi-Primary Mode

In multi-primary mode, all nodes accept writes simultaneously. Conflict detection handles concurrent updates to the same rows.

dbdeployer deploy replication 8.4.8 --topology=group

All nodes are writable:

~/sandboxes/group_msb_8_4_8/n1 -e "CREATE DATABASE test1"
~/sandboxes/group_msb_8_4_8/n2 -e "CREATE DATABASE test2"
~/sandboxes/group_msb_8_4_8/n3 -e "SELECT schema_name FROM information_schema.schemata"

Monitoring: check_nodes

The check_nodes script queries performance_schema.replication_group_members on each node and summarizes the group state:

~/sandboxes/group_msb_8_4_8/check_nodes
# node 1 - ONLINE (PRIMARY)
# node 2 - ONLINE (SECONDARY)
# node 3 - ONLINE (SECONDARY)

Available Scripts

Script	Purpose
`n1`, `n2`, `n3`	Connect to node 1, 2, 3
`check_nodes`	Show group membership and role of each node
`start_all`	Start all nodes
`stop_all`	Stop all nodes
`use_all`	Run a query on all nodes
`test_replication`	Verify data propagates across all nodes

Controlling the Number of Nodes

Use --nodes to deploy more than three nodes:

dbdeployer deploy replication 8.4.8 --topology=group --nodes=5

Concurrent Deployment

Large clusters start faster with --concurrent:

dbdeployer deploy replication 8.4.8 --topology=group --nodes=5 --concurrent

InnoDB Cluster: the Managed Alternative

MySQL InnoDB Cluster wraps Group Replication with MySQL Shell (for orchestration) and MySQL Router (for transparent failover routing). If you need the full managed stack, see InnoDB Cluster.

For plain Group Replication without the Shell/Router overhead, the --topology=group approach on this page is sufficient.