2017-04-13 08:18 AM
We are doing 7 mode fabric metrocluster tests on a customer. In almost all the tests we did unti now , have no problem, but in the scenario below, we see that the VMs in the ESXi hosts lost datastores when we shutdown the shelf on that with existing root aggregate of site-A. Do I have to miror root-aggregate on Site-A? Do I have to do this?
Expected behaviour: Relevant disks for offline and the plex is broken. There’s no disruption to data availability to hosts running HA (VMware High Availability) or FT (Fault Tolerance), no change is detected by the ESXi Server. When the shelf is powered back on the plexes will sync automatically
Solved! SEE THE SOLUTION
2017-04-13 08:27 AM - edited 2017-04-13 08:30 AM
Yes, root aggregates have to be mirrored.
If you check the messages file you will even see that ONTAP is complaining about the root aggregate not being mirrored.
In that failure case the controller would just access the mirror plex and continue to serve all data.
2017-04-13 08:51 AM
Thanks for quick reply.When I open the NetApp case and ask for 7 mode fabric metrocluster they said that it is not necessary(must) to mirror the root aggregate. Do you mean "have to" for my senario? Or do you always need to mirror it?
2017-04-13 10:12 AM
In order for that particular failure scenario to work - yes, you have to mirror the root aggregate.
It's generally required to mirror the root aggregate to be fully protected.
For active/passive configurations I have seen people not mirror the root aggregate for the passive controller in order to save on disks, but ONTAP will complain.
In order to have a fully supported and non-complaining system, the root aggregates must be mirrored.