AFM DR use case

This use case describes the AFM-based Asynchronous Disaster Recovery (AFM DR) process when NFS and SMB clients are mounted from the primary IBM Storage Scale cluster.

Overview of AFM DR when NFS and SMB clients are mounted from the primary cluster

During scheduled maintenance on the primary cluster or a failure at the primary site cluster, those NFS, and SMB clients that mount exports from the primary cluster experience a brief and temporary outage, when the mmafmctl failover commands are run. After the mmafmctl failover commands complete, the NFS and SMB clients can mount from the new acting primary cluster (formerly the secondary cluster).

Note: To ensure a successful failover, system administers at each site need to create identical export definitions on the secondary cluster. It is important to keep the definitions up to date when changes occur on the primary cluster.
After you create initial AFM DR relationships, you can create NFS and SMB exports so that applications can access the IBM Storage Scale data via protocols.

Use case 1: The path to the data is identical on both the primary and secondary clusters

  1. On both the primary and the secondary clusters, enter the same path and client (-c) definitions as shown in the following example:
    mmnfs export add /ibm/gpfs0/export1 -c
9.11.102.0/24(Access_Type=RO,Squash=root_squash;
9.11.136.0/23(Access_Type=RO,Squash=root_squash;
10.18.40.0/22(Access_Type=RW,Squash=root_squash;
10.18 48.0/22(Access_Type=RW,Squash=root_squash;
  2. On the NFS clients, run the following commands.
    • Issue the showmount -e command and the IP address of the primary cluster as shown in the following example:
      showmount -e 10.18.96.100
Export list for 10.18.96.100:
/ibm/gpfs0/export1
9.11.102.0,911.136.0,10.1840.0,10.18.48.0
    • Issue the showmount -e command and the IP address of the secondary cluster as shown in the following example:
      showmount -e 10.18.50.30
Export list for 10.18.50.30:
/ibm/gpfs0/export1
9.11.102.0,911.136.0,10.18.40.0,10.18.48.0
  3. On the NFS clients, issue the following command to mount from the primary CES IP addresses as shown in the following example:
    mount -o hard, sync
10/18.96.100:/ibm/gpfs0/export0 /mnt_pnt
  4. If a failure occurs on the primary cluster, initiate failover.
  5. On the NFS clients, issue the following commands to unmount from the primary CES IP addresses and then mount from the secondary CES IP addresses as shown in the following example:
    umount -f /mnt_pnt
mount -o hard, sync
10.18.50.30:/ibm/gpfs0/export0 /mnt_pnt

Use case 2: The path to the data is different on the primary and secondary cluster

  1. Create the export with the same client (-c) definitions.
    • Issue the following command and the IP address of the primary cluster:
      mmnfs export add /ibm/gpfs0/export1 -c
9.11.102.0/24(Access_Type=RO,Squash=root_squash;
9.11.136.0/23(Access_Type=RO,Squash=root_squash;
10.18.40.0/22(Access_Type=RW,Squash=root_squash;
10.18 48.0/22(Access_Type=RW,Squash=root_squash;
    • Issue the following command and the IP address of the secondary cluster:
      mmnfs export add /ibm/gpfs3/export9 -c
9.11.102.0/24(Access_Type=RO,Squash=root_squash;
9.11.136.0/23(Access_Type=RO,Squash=root_squash;
10.18.40.0/22(Access_Type=RW,Squash=root_squash;
10.18.48.0/22(Access Type=RW,_quash=root_squash;
  2. On the NFS clients, run the following commands.
    • Issue the showmount -e command and the IP address of the primary cluster:
      showmount -e 10.18.96.100
Export list for 10.18.96.100:
/ibm/gpfs0/export1
9.11.102.0,911.136.0,10.18.40.0,10.18.48.0
    • Issue the showmount -e command and the IP address of the secondary cluster:
      showmount -e 10.18.50.30
Export list for 10.18.50.30:
/ibm/gpfs3/export9
9.11.102.0,911.136.0,10.18.40.0,10.18.48.0
  3. On the NFS clients, issue the following command to mount from the primary CES IP addresses:
    mount -o hard, sync
10/18.96.100:/ibm/gpfs0/export0 /mnt_pnt
  4. On the NFS clients, issue the following commands to unmount from the primary CES IP addresses and then mount from the secondary CES IP addresses:
    umount -f /mnt_pnt
mount -o hard, sync
10.18.50.30:/ibm/gpfs3/export9 /mnt_pnt