Hello, we want to deploy a select metrocluster;
we already did some tests and we saw that when one node and the mediator fails, the surviving node doesn't serve its svm-lifs anymore. even when the mediator is back the svm doesn't take the lif online. is this an expected behaviour? any ideas on this?
#Ontap Select Metrocluster
pallid trail
queen ingot
there is no "ONTAP Select MetroCluster". MetroCluster is a Hardware thing.
But to answer your question, yes, if you have multiple failures, the failover will not work correctly. You generally only protect against single failures, not multiple concurrent failures. In your case you have to manually intervene
pallid trail
Hi @queen ingot !
thank you for your response.
I'm refering to ONTAP Select MetroCluster SDS https://docs.netapp.com/us-en/ontap-select/concept_usecase_mc_sds.html
Sure that multiple failures need manual work to do. That the failover doesn't work in this scenario with double failures that's is pretty clear. I just thought that the surviving site continues its own workload (Lifs) but it reacts like taking things offline when isolated: which is okay.
where I am surprised is that even when the mediator is back the node still doesn't take any of its lifs online.
queen ingot
ah right, sorry, they renamed this to MetroCluster... I'm not an expert with Select SDS but I agree that you would expect things to settle back down when the mediator returns. Maybe someone else can shed some light on this
pliant agate
the mediator isn't really useful if it can't observe the failure and initiate site failover... it can't really usefully be on either of the esx clusters where the mcc is running.
fickle halo
regarding the mediator, It should not re-use hardware/infrastructure that might be effected by site A or Site B failure as much as possible. It needs to be 3rd site, essentially.
restive coral
I wish we supported it on armhf.. had a few customers ask for it 😉