High Availability and Failover
1. High Availability and Failover
The former 2019 MariaDB/CentOS design essay is obsolete. Detailed production topology and failover runbooks must remain restricted.
1.1. Current architectural entry points
modules/ProvHA— HA-specific commands, observers, and slave configuration rebuildsdatabase/migrations/BaseMigration.php— database/system scope and HA-aware behaviorapp/Console/Kernel.php— module-aware schedulingqueue worker and service configuration under
Install/files/
1.2. Public-safe principles
Separate stateless application processes from stateful databases/services
Make config rebuild and synchronization commands idempotent
Use read-only or role-specific connections where required
Prevent overlapping scheduled rebuilds
Test failover in an approved non-production environment
Exact node roles, addresses, credentials, and recovery procedures belong in restricted maintainer documentation.