shell walrus
ONTAP 9.14.1P6, 4 x AFF-A300 controllers, Atto Bridges which does not support ONTAP 9.15>
I have found this documentation:
https://kb.netapp.com/on-prem/ontap/mc/MC-KBs/Converting_a_4-Node_Fabric_MetroCluster_to_Non-MetroCluster_HA_preserving_data
My question - is it possible to preserve the data on the mirrored half of the disks, by un-assigning the remote copy of the disks, in order to re-assign that disks to the local HA pair?
novel crater
Don't think so... at least not if you just do a "aggr mirror break" kind of command... because then the broken off disks will end up as spares... So you would end up in a kind of forcefully break of the mirror in a split-brain simulated situation where you can force a switchover on each side... but pretty sure it's not supported... An alterntaive way would be to break the mirror and the MC, then setup a new cluster using the new spare disks, then snapmirror over the volumes you need on that system... And about the ATTO bridges I am pretty sure it will work with 9.15 without any issues... (maybe not officially supported)... you can buy used "newer" ATTO bridges for not much money and just replace them one by one...
shell walrus
Thanks for the feedback. Yes - that was what I thought as well -to mirror the data to the second DC.
shell walrus
Another idea that just popped up. What if I un-assign the data disks from the primary controller in the one DC, and assign it to the partner controller in the second DC (physical move it?)
novel crater
Might work, but not supported at all... I haven't actually tried it on cdot... on 7mode it worked every time... generally you can get away with many things, but when you get into unsupported stuff I would not do it with data that you cannot afford to loose... :.-)
...and you have to bring the whole cluster down, boot one node into maint mode, and forcefully remove ownership on the specific disks, then assign them to the other controller... might work, but don't come crying when you have lost your data 😉
...and the aggregate might not just appair automatically... you will most likely have to run some cluster database command in order to import it... so basically high risk move...
shell walrus
What about I unjoin the one node from the one cluster and join it again on the other side.
cunning peak
you cannot split a MetroCluster into two HA pairs. The metrocluster unconfigure command only works after you have deleted all SVMs (and thus all volumes) from both sides of the MetroCluster
shell walrus
According to the document I mentioned above you can, and it is supported.
= preserving data = no data loss?
cunning peak
hm. I tried this once and it didn't work. But that was an IP metro and it was a few ONTAP versions back. Maybe something changed since then 🤷♂️
shell walrus
I will in any case open a support case before we start this, and run it by the Technical Support Team.