Administrator's Guide Supporting RSA Data Protection Manager (DPM) Environments (Supporting Fabric OS v7.2.0) Owner's manual

Fabric OS Encryption Administrator’s Guide (DPM) 257
53-1002922-01
Encryption group merge and split use cases
6
Recovery
If auto failback policy is set, no intervention is required. After the node has come back up, all
devices and associated configurations and services that failed over earlier to N1 fail back to N3.
The node resumes its normal function.
If auto failback policy is not set, invoke a manual failback if required. Refer to the section
“Performing a manual failback of an encryption engine” on page 254 for instructions.
A member node lost connection to the group leader
AssumeN1, N2 and N3 form an encryption group, and N2 is the group leader node. N3 and N1 are
part of an HA cluster. Assume that N3 lost connection to the group leader node N2 but still
maintains communications with other nodes in the encryption group.
Impact
Failover to N1 does not occur, because the isolated node and the encryption engines’ encryption
services continue to function normally. However the disconnect of N3 from the group leader breaks
the HA cluster and failover capability between N3 and N1.
You cannot configure any CryptoTargets, LUN policies, tape pools, or security parameters that
would require communication with the isolated member node. In addition, you cannot start any
rekey operations (auto or manual).
Refer to the section “Configuration impact of encryption group split or node isolation” on page 264
for more information on which configuration changes are allowed.
Recovery
Restore connectivity between the isolated node and the group leader. No further intervention is
required.
A member node lost connection to all other nodes in the encryption
group
Assume N1, N2 and N3 form an encryption group and N2 is the group leader node. N3 and N1 are
part of an HA cluster. Assume that N3 lost connection with all other nodes in the group. Node N3
finds itself isolated from the encryption group and, following the group leader succession protocol,
elects itself as group leader. This action splits the encryption group into two encryption group
islands. EG1 includes the original encryption group minus the member node N3 that lost
connection to the encryption group. EG2 consists of a single node N3, which functions as the group
leader. Both EG1 and EG2 are in a degraded state.
Impact
• The two encryption group islands keep functioning independently of each other as far as host
I/O encryption traffic is concerned.
• Each encryption group registers the missing members as “offline”.