To have automatic failover between controllers both must be up and running. To run controller needs to have at least root volume – i.e. disks. So if you assign all disks to one controller second won’t run and you will lose redundancy – i.e. controller failover, possibility to do NDU.
You don't need a spare to run the controller, that's correct...but this is the opposite of NetApp's best practices. What happens when you loose a drive on a 2 disk raid4? What did you do about the endless messages that pop up about not having a spare available? "I do not recommend it, but it is doable" should have been the first piece of advice given if you're leading someone down the wrong path to start.
raid 4 has other issues in regards to performing NDU's also, makes it a pain when you want to upgrade to a later version of ONTAP...
personally, I'd use three disks in an aggregate, don't use a spare (unless a fourth disks was available, defo recomended to have a spare) and the turn the raid.min_spares option (or something like that) to 0 to stop the messages...
Another consideration is raid.timeout. When there is a disk failure, and there is no spare available (or if the nvram battery fails), raid.timeout (default = 24 hours) goes into effect and the system will shutdown. You can modify this option or replace the failed disk with spare before the timeout (assuming not a weekend with no one around) but not best practice as mentioned earlier. Also, rapid raid recovery and maintenance garage are not available with no spares. If a passive node less of an issue, but maybe go to the pencil fund and buy an extra disk drive
What is the big concern about not having a hotspare if the Cluster node is used for Failover only, Active / "Passive"? I would not put any user data on the passive side and could even snapmirror the root volume to the Active controller or do an NDMP backup.
Even if the Passive node shuts down one would only then be exposed to a panic or someother really bad event on the "Active" Side during that outage. This seem like a pretty low risk.
It would be nice if Netapp gave you an option to run Ontap off of an SD Card or a USB drive... and let allthe Disks belong to the Active Node. I run production vShpere 4i servers installed on USB drives an SD cards why not Ontap?
Hello everyone,hope you canresolve mydoubts.Just bought aFAS2040C, which has 2controlleractive / active. Thefilercomes with 6disks in thechassis and by default 3 diskshave been distributedtoeach controller.My intentionis to usethe 6diskson a controller,leaving the otherwithoutdiscs, sothat would be turn off until a few months, when I connectto an expansion unit byFC.
The question is whethereliminating theaggregated that contains theroot volume, Ii will be able to release the disks toassign them to the other controller, and do not know ifI am destroyingvital informationonthe second controllerto activate it in next months.By eliminating therootvolume, I guessI'mdestroying theSO (DataOntap).Am Iright?, How should proceed?