Netapp cluster failover

[stevenzhang]

Hi experts

We plan to create one cluster with two HA pairs, each pair has two nodes,the cluster provide nas service only(nfs ).

Is it ok to realize HA that  one ha pairs fails (two node in the pair fail), there is no problem with client nfs access.

If it is , how to config HA?

[aborzenkov]

If both nodes in HA pair fail, clients will lose access to data on disks, connected to this HA pair. Also clients will lose connectivity to cluster if LIF they use cannot failover to another HA pair.

[stevenzhang]

Thanks for help.

Q1>Configure two ha pairs to ontap cluster, when one pair fails, will  lif failover to another ha pair ?

Q2>if use external storage as data disk with FC swith(don't use disk shelf), is it ok to map the same lun to four nodes?when two nodes in one pair fail, can the other pair  still access the lun?

[aborzenkov]

Q1. That depends entirely on your configuration - failover-group and failover-policy, assigned to LIF.

Q2. No, that’s not possible.

[FMS]

Q1 : A cluster has 4 nodes consisting of HA1 and HA2 out of which  HA1 goes down (both controllers fails  but SP is up), Will HA2 be able to serve the data of the aggregate created on HA1 if portsets has been created of consuming the lif of both the HA?

What I think is the Answer to the question is "no" but according to the link below it is possible.

Please explain?

Quoting the information from the link provided.

"Your data is still up and good on HA2, but by limiting the access to a portset that only uses ports from HA1, you have lost access to the data, because HA1 is still the gatekeeper to that igroup and LUN."

https://community.netapp.com/t5/Data-ONTAP-Discussions/Moving-a-volume-to-an-aggregate-on-the-second-HA-within-a-4-node-CDOT/td-p/111139

[SpindleNinja]

You also need to keep in mind the idea of epsilon and quorum with a 4 node cluster.  In a 4 node cluster, if 2 nodes go down, there are not enough nodes left in the cluster to quorum and the remaining two shut down data services.      in a 6 node for example,  if 2 fail,  the remaining 4 nodes will function OK.   

[FMS]

So, in a 6 node cluster, an aggregte created on HA1 will still be available if HA1 goes down and 4 nodes are still available maintaining quorum and serving the data from HA2 and HA3 even though the disk connected only with HA1.

[aborzenkov]

---

@FMS wrote:

So, in a 6 node cluster, an aggregte created on HA1 will still be available if HA1 goes down

---

Is this a question (there are no question marks in your post)? Then the answer is “no”.  If this is a statement, it is incorrect, sorry.

[SpindleNinja]

Everything local to the HA pair will be offline when the pair goes down.   

If you need to protect agaist HA pair failure or site failure, you can add extra resiliency by installing a Metrocluster.