We are dealing with a complicated upgrade from a stretch metrocluster to a fabric metrocluster.
The stretch cluster contains currently multipe EXN2000 (ds14mk2) and EXN4000 (ds14mk4), the controller is recently upgraded to a N6240 (FAS3240) with ONTAP 8.01P3
We want to replace the old EXN2000 (ds14mk2) with new SAS EXN3000 (ds4243) and convert to a fabric MetroCluster.
The challenge is to maintain service as long as possible and endure the lowest amount of downtime (6 hours max).
To make matters more interesting, we do not have the available space left to migrate storage from the existing EXN2000 and to the EXN4000.
The vicious circle is the need for ONTAP 8.1.1 and higher to use the ATTO fibrebridges in a fabric MetroCluster, but as you might know these do not support ESH-2 modules anymore. Although NetApp does support the upgrade from ESH-2 to ESH-4 modules in a ds14mk2 shelf, IBM does not in a EXN2000 shelf.
IBM requests an RPQ for testing this situation. Alternatives are to use temporary extra EXN4000 shelves and migrate the EXN2000 data to the EXN4000 and then during a issue a MetroCluster failover and disconnect theses shelves.
Another alternative is to replace the non-support ESH-2 modules with ESH-4, but this can take allot of time before a RPQ request is being testen and confirmed. The last option considered is replacing the shelves with EXN4000 and use the same disks and then migrate to the EXN3000 en remove the temporary EXN4000's.
Does anyone have experience with this issue or maybe performed such a migration.
It seems unlikely that this case is rare, because of the end-of-support of the ESH-2 modules and the convergence to SAS shelves.
The last question is: Does NetApp have a manual or whitepaper for upgrading from stretch to fabric metrocluster. I recall seeing it, but i could not find it anymore. Maybe I am wrong.