Ability for protocol replication to co-exist with other non-protocol AFM-based DR instances

We need protocol replication to be able to co-exist with other non-protocol AFM-based DR instances

On our advanced edition GPFS clusters, we wish to have both:

o AFM-based DR for normal filesets (managed via mmafmctl)

o protocol DR for filesets with NFS or SMB exports (managed via mmcesdr)

running between the same pair of clusters. This is not supported. Indeed, in testing we have found that:

1) executing “mmcesdr secondary failover” causes ALL the AFM filesets at the secondary site to be converted from secondary to primary AFM filesets. It also fiddled with their NFS exports at the secondary site.

2) When mmcesdr is used to execute the failback to the old primary, the “protocols” filesets (those with exports and managed by mmcesdr) correctly failback and replication for these continues. However, the non-protocols filesets that were converted by “mmcesdr secondary failover” to primary fileset are NOT converted back to AFM mode primary when an mmcesdr failback process is executed.

If mmcesdr is going to failover non-protocol AFM filesets, it also needs to manage their failback properly. Or not touch the non-protocol AFM filesets at all.

As it stand one has to go and manually fix up the messed up AFM setup for ALL non-protocol replication the at both the primary and secondary sites, as well as fix the modified NFS exports at the secondary site.