LIF replication and placement requirements and issues
You should be aware of the LIF replication requirements in a MetroCluster configuration. You should also know how a replicated LIF is placed on a partner cluster, and you should be aware of the issues that occur when LIF replication or LIF placement fails.
Replication of LIFs to the partner cluster
When you create a LIF on a cluster in a MetroCluster configuration, the LIF is replicated on the partner cluster. LIFs are not placed on a one-to-one name basis. For availability of LIFs after a switchover operation, the LIF placement process verifies that the ports are able to host the LIF based on reachability and port attribute checks.
The system must meet the following conditions to place the replicated LIFs on the partner cluster:
| Condition | LIF type: FC | LIF type: IP/iSCSI |
|---|---|---|
| Node identification | ONTAP attempts to place the replicated LIF on the disaster recovery (DR) partner of the node on which it was created. If the DR partner is unavailable, the DR auxiliary partner is used for placement. | ONTAP attempts to place the replicated LIF on the DR partner of the node on which it was created. If the DR partner is unavailable, the DR auxiliary partner is used for placement. |
| Port identification | ONTAP identifies the connected FC target ports on the DR cluster. | The ports on the DR cluster that are in the same IPspace as the source LIF are selected for a reachability check. If there are no ports in the DR cluster in the same IPspace, the LIF cannot be placed. All of the ports in the DR cluster that are already hosting a LIF in the same IPspace and subnet are automatically marked as reachable; and can be used for placement. These ports are not included in the reachability check. |
| Reachability check | Reachability is determined by checking for the connectivity of the source fabric WWN on the ports in the DR cluster. If the same fabric is not present at the DR site, the LIF is placed on a random port on the DR partner. | Reachability is determined by the response to an Address Resolution Protocol (ARP) broadcast from each previously identified port on the DR cluster to the source IP address of the LIF to be placed. For reachability checks to succeed, ARP broadcasts must be allowed between the two clusters. Each port that receives a response from the source LIF will be marked as possible for placement. |
| Port selection | ONTAP categorizes the ports based on attributes such as adapter type and speed, and then selects the ports with matching attributes. If no ports with matching attributes are found, the LIF is placed on a random connected port on the DR partner. | From the ports that are marked as reachable during the reachability check, ONTAP prefers ports that are in the broadcast domain that is associated with the subnet of the LIF. If there are no network ports available on the DR cluster that are in the broadcast domain that is associated with the subnet of the LIF, then ONTAP selects ports that have reachability to the source LIF. If there are no ports with reachability to the source LIF, a port is selected from the broadcast domain that is associated with the subnet of the source LIF, and if no such broadcast domain exists, a random port is selected. ONTAP categorizes the ports based on attributes such as adapter type, interface type, and speed, and then selects the ports with matching attributes. |
| LIF placement | From the reachable ports, ONTAP selects the least loaded port for placement. | From the selected ports, ONTAP selects the least loaded port for placement. |
Placement of replicated LIFs when the DR partner node is down
When an iSCSI or FC LIF is created on a node whose DR partner has been taken over, the replicated LIF is placed on the DR auxiliary partner node. After a subsequent giveback operation, the LIFs are not automatically moved to the DR partner. This can lead to LIFs being concentrated on a single node in the partner cluster. During a MetroCluster switchover operation, subsequent attempts to map LUNs belonging to the storage virtual machine (SVM) fail.
You should run the metrocluster check lif show command after a takeover operation or giveback operation to verify that the LIF placement is correct. If errors exist, you can run the metrocluster check lif repair-placement command to resolve the issues.
LIF placement errors
LIF placement errors that are displayed by the metrocluster check lif show command are retained after a switchover operation. If the network interface modify, network interface rename, or network interface delete command is issued for a LIF with a placement error, the error is removed and does not appear in the output of the metrocluster check lif show command.
LIF replication failure
You can also check whether LIF replication was successful by using the metrocluster check lif show command. An EMS message is displayed if LIF replication fails.
You can correct a replication failure by running the metrocluster check lif repair-placement command for any LIF that fails to find a correct port. You should resolve any LIF replication failures as soon as possible to verify the availability of LIF during a MetroCluster switchover operation.