Understanding and Designing Serviceguard Disaster Recovery Architectures

In addition, notifications are sent to the file /var/opt/resmon/log/cc/eventlog on the
cluster node where monitoring package is running.
Types of Cluster Events
Cluster events are of two types in Continentalclusters, Alerts and Alarms. At least one alarm event
must be triggered for the recovery process to start normally. If there are no cluster events, the
recovery process cannot start in Continentalclusters.
Cluster event of type alerts are typically used for informational purposes. For example, if the user
wants to be notified when the primary cluster goes down, a cluster alert event for status down can
be defined with a time duration of 0 seconds to ensure instantaneous notification if the primary
cluster goes down.
Cluster event of type alarms are typically used to indicate failures. For example, if a cluster has
remained unreachable for 10 minutes, it could mean a failure. A Cluster Alarm event must occur
to enable recovery on the recovery cluster.
However, a Cluster Alarm or Alert configured for the cluster status being UP will not enable recovery.
Manual Recovery
There are some cases where the cluster health cannot be determined because the monitoring
happens over a network. For example, the cluster status can be reported as unreachable when
either the network service has failed or when the cluster nodes are rebooting. Hence, information
needs to be gathered from an independent source before proceeding with the recovery process.
It is for this reason the recovery in Continentalclusters is manually initiated but automatically
recovered.
After a cluster event of type alarm is triggered, the recovery process can be started using a single
command. The user can choose to recover all the recovery groups or can specify the names of the
recovery groups that need to be recovered.
Types of Configuration
Continentalclusters support the following configurations:
1. Basic configuration
2. Bi-directional/Mutual Recovery configuration
3. N-1 Configuration
4. N-1 Bidirectional configuration
5. Continentalclusters Cascading Failover
Basic configuration
In a basic Continentalclusters configuration, there is one primary cluster and one recovery cluster.
This type of a configuration is ideal when there is only one cluster running production applications
and the other cluster is on a standby mode.The standby cluster can become the production cluster
when there is a disaster.
Understanding Continentalclusters 29