Skip to main content

HA Deployment

An HA deployment is a production environment run as an active / warm-standby pair: two Linkiir servers sharing one working directory and one log database, with a front door sending clients to whichever is active. If the active server fails, the standby promotes itself in about half a minute and no messages are lost.

How it differs from a single-server PROD environment

PRODHA
Linkiir servers12, one active at a time
Working directoryLocalShared, one directory both servers open
Log databasePostgreSQL or MS SQLPostgreSQL or MS SQL, reachable from both
BrokerExternal clusterExternal cluster, three nodes
Server failureA maintenance eventA failover of seconds
LicenseEnterpriseEnterprise with the HA feature

Everything else is the same: the same release, the same projects, the same import and export path between environments.

Full documentation

The High Availability section covers it in depth:

PageCovers
High AvailabilityWhat HA protects against, and what it does not
TerminologyActive, standby, failover, failback, quorum, RPO and RTO
LicensingOne Enterprise license with the HA feature per pair
ArchitectureThe design, key components, and the failover sequence
System RequirementsSizing, shared storage, database, broker, ports
TopologiesThe five supported deployment shapes
Using the HA SettingsThe High Availability page: enable HA, name the servers, step down
Operating an HA PairTesting a failover, patching without downtime, monitoring
Backup and DRWhy HA is not a backup, and how DR differs
Planning Your DeploymentWhat to decide and prepare before the build
HA is not a backup

Both servers read one copy of your data, so a deletion or a bad change affects both at once. An HA deployment still needs instance backups, a remote per project, and database backups. See Backup and Restore.