Data Consistency for HA
When an HA failover occurs, JSA ensures the consistency of your data.
The type of storage that you use determines how HA data consistency is maintained. If you configure HA with external storage, data consistency is maintained by using a component such as an iSCSI external storage device. See Offboard Storage Requirements for HA.
If you do not use external storage devices, then JSA HA maintains data consistency between a primary and secondary HA host by using Distributed Replicated Block Device (DRBD).
Data synchronization occurs in the following situations in an HA environment:
When you initially configure an HA cluster.
When a primary HA host is restored after a failover.
During normal HA operation, data is synchronized in real time between the primary and secondary host.
Real-time Data Synchronization
When you configure an HA cluster, the /store file system on the primary HA host is automatically synchronized with the /store partition on the secondary HA host.
If the primary HA host fails over, the /store file system on the secondary HA host is automatically mounted to its local disk, where it continues to read from and write to the data received by the primary HA host before the failover.
After synchronization is complete, the secondary HA host assumes a status of standby.
Depending on the size of the primary /store partition and performance, disk synchronization can take an extended time period. Ensure that the connection between the primary and secondary HA host has a minimum bandwidth of 1 Gbps.
Post-failover Data Synchronization
Data that is collected by a primary high-availability (HA) host, up to the point of failover, is maintained virtually, in real time, by the secondary HA host.
When the primary HA host is restored after a failure, only the data that is collected by the secondary HA host in the intervening period is synchronized with the primary HA host. Therefore, post-failover disk synchronization is faster than initial disk synchronization, unless the disk on the primary HA host was replaced or reformatted when the host was manually repaired.
When restored from a failover, the status of the primary HA host becomes offline. You must set the primary HA host to an online state, and set the secondary host to an offline state, before it can become the active host. Disk replication with the secondary HA host is enabled while the primary HA host remains offline.