Example: One cluster which consists of two FAS2750HA (4 nodes in total), both FAS2750HA systems have two DS460C attached. Disks in head are only for boot aggregate.
Would it be possible to migrate this into a 2 node cluster with a FAS8200/8300 model?
We have spare disks and shelf so that we could add the FAS8200/8300 into the cluster. But is there a way to migrate the aggregates between HA-pairs? It doesn't need to be non-disruptive, we are allowed down time if we need it...
#Aggregate migration with data?
uncut swift
...A first "headswap" of one of the FAS2700 to FAS8200/8300 would be pretty easy, but it's the part of somehow migrating the aggregates from the two other nodes in the other HA-pair that is the tricky part 😉 If it was 7-mode ther would be no problems, but I have never tried to just attach the disks and assign them from another ha-pair... my guess is that it would most likely work? But most likely not supported like this? It would be a pain if we had to migrate via volume moves... 120 16TB disks which are more than 70% filled... that's a bit of moving... and we need some kind of "switch gear" to pull it off....
shy iron
The "Upgrade by moving storage" workflow should fit: https://docs.netapp.com/us-en/ontap-systems-upgrade/upgrade/upgrade-decide-to-use-this-guide.html
you need to reassign the disk ownership
uncut swift
The "Upgrade by moving storage" workflow should fit: https://docs.netapp.com/us-...
Yeah that describes the "headswap" of one of the FAS2700 HA-pairs, but can we just shutdown the other HA-pair, connect their shelfs to the new cotroller which already have the other shelfs connected? I guess an easier way to say it would be migrate 2 x HA-Pairs into 1 x HA-Pair...
shy iron
If you reassign the disk ownership to the sys-ids of the new nodes that should work if I'm not misunderstanding something here.
naive wedge
we had this discussion somewhere on the discord in the past. i.e. relocating an aggregate from one HA pair to another in the same cluster. I think the outcome was that it should work (with some extensive cleanup afterwards) but it was not a supported procedure...
mint leaf
Correct, it is not supported to merge aggregates from multiple clusters. IIRC PS has a procedure which will result in a supported system. It does require a LOT of cleanup, since the volumes don't appear in the VLDB and aren't owned by any valid SVM
uncut swift
Correct, it is not supported to merge aggregates from multiple clusters. IIRC PS...
I can understand the issue if you add an aggregate from another cluster which will result in issues becuase the ownership-information of the volumes are lost, makes sense. But if it's from the same cluster all the ownership should be OK. So a basic procedure would be to move the SVM's root volume to the "surviving" HA-Pair's aggregate and of cause move the LIFs as well... and then shutdown the HA-Cluster and re-cable the disks and do the re-assign of the disks.. Not sure if this is possible while the cluster is up, or if we need to bring both nodes down into maint-mode.? thankfully this is "just" backup data, so downtime isn't a big problem... The main reason we would like this is that the FAS2700 has a max of 144 disks and we have two DS460C on each HA-pair... we could also just add another FAS2700-ha-pair into the cluster... we have been talking alot about which approach would be the best... mode ha-pairs in one cluster, or one more powerfull ha-pair (ie. fas8200/8300)... so we are trying to describe both approaches to management, and what is possible in the future...