Bright:HA concepts

What is HA

A cluster with a single head node has a single point a failure. In other words if the head node fails the entire cluster will be down. HA allows a cluster to have two head nodes to allow fail-over.

HA concepts

Each of the head nodes will have two states associated with it - Primary or Secondary, Active or Passive. Under normal circumstances the Primary head node will be in active mode, while the secondary head ndoe will be in passive mode.

Services running on the head node

These service will be running on the active head node:

• CMDaemon: providing certain functionality on both head nodes (e.g. provisioning)
• DHCP: load balanced setup
• LDAP: running in replication mode (the active head node LDAP database is pulled by the passive)
• MySQL: running in master-slave replication mode (the active head node MySQL database is pulled by the passive)
• NTP
• DNS

Several services will migrate on failover between the head nodes:

• NFS
• The User Portal
• Workload Management (e.g. SGE, Torque/Maui)