Usage
This command manages Assure MIMIX for AIX's disaster recovery processes as well as failover and failback operations. Given a primary context <X> configured on both a “Production” and a “Recovery” Server, note:
-
To create an associated failover context <Y>, on both the production and recovery servers execute:
rtdr -C <X> -F <Y> setup-
To failover to the Recovery server, on the recovery server execute:
rtdr -C <X> failover-
To re-synchronize the revived production server, on both the production and recovery servers execute:
rtdr -C <X> resync-
To failback to the production server, first on the recovery server, and then on the production server execute:
rtdr -C <X> failback-
To select and configure from one of many recovery servers (a cascaded replication configuration) as the target machine for all subsequent failover, resync and failback operations execute:
rtdr -C <X> -F <Y> -s <hostname> setup-
To create and setup a failover context, on both the production and recovery servers:
rtdr -C <X> -F <Y> setup-
Prior to failover, you should validate the data integrity of the Replica, and if necessary, validate the data if necessary.
To validate data integrity of the Replica, create a snapshot of the replica, then analyze it with the application itself. To create the snapshot, on the recovery server:
scrt_ra -C <X> -X
-
If your analysis indicates data corruption, use a virtual restore to locate and validate an optimal restore point. A virtual restore leaves a snapshot in place for analysis at the restored point in time. To perform a virtual restore, on the recovery server:
scrt_ra -C <X> [-D | -S | -t]-
Given a corrupt replica, and a validated restore point in time, use a failover restore to roll the actual data replica, not a snapshot of the replica, back to the validated point in time. To perform a failover restore, on the recovery server execute:
scrt_ra -C <X> -F [-D | -S | -t]After failover, start the application on the recovery server. It will be the acting production environment until failback. All data modifications will be tracked and shipped back to the original production server by resync.
After reviving the original production server, use resync to bring the production server data up to date with the recovery server data.
After the resync completes, use failback to return to the original production and recovery server roles. Both the production and recovery servers must be live to execute resync and failback.
-
To dirty all statemaps, in the failover context on the recovery server (the acting production server):
rtstop -C <X> -F
scconfig -C <X> -M After validating the replica, the disaster recovery procedure is to failover, resync, then failback.
After failover, start the application on the recovery server. It will be the acting production environment until failback. All data modifications will be tracked and shipped back to the original production server by resync.
After reviving the original production server, use resync to bring the production server data up to date with the recovery server data.
After resync completes, use failback to return to the original production and recovery server roles. Both the production and recovery servers must be live to execute resync and failback.
-
To dirty all statemaps, in the failover context on the recovery server (the acting production server):
rtstop -C <X> -F
scconfig -C <X> -M
rtstart -C <X>After a system failover, Assure MIMIX for AIX cannot rollback to a point in time before the failover. Likewise, after a system failback, Assure MIMIX for AIX cannot rollback to a point in time before the failback. For this reason, replica data integrity should be validated with a snapshot prior to executing a failover.
Syntax
rtdr -C <ID> [-fhmnqv] failover | resync | failback
rtdr -C <ID> -F <ID> [-s <hostname>] [-fhmnv] setup
-C Context ID (of the "primary" context)
-F Failover Context ID
-f Forced execution (use with caution)
-h Help, prints usage
-m Man style help
-n No execution, just print commands
-q quiet, do not ask for confirmation
-s Select failover site server from multiple recovery
servers (default is first replication hop's server.)
<hostname> must be a configured SCRT/info/host HostName attribute.
-v Verbose output