ONTAP Hardware

FAS2600 with ONTAP 9.2: failover to other node if all data network paths to the first node fail?

nhwanderer
2,854 Views

I'm trying to design the network topology for our new FAS2620 two node HA pair running 9.2. At this point it's likely we'll run active/passive. 

 

My intention is to to create a LACP network pair on each node for the data path, and put the primary node on one switch cluster, and the secondary node on another switch cluster.

 

The switch clusters are Aruba/HP ProCurve 2920 systems, where the only true single point of failure is the switch stack software, OR when I update the switch stack code, and that takes about three minutes, which brings the stack down during that time.

 

With the LIF and node failover policies available to us in Ontap 9.2, if the switch stack connected to the primary node goes down resulting in a complete lack of connectivity to the data path of the primary node, is there a way to get the passive secondary node to take automatically take over in some way, either through full failover, or just grabbing the lif?

 

If this isn't possible with active/passive, is it possible with active/active?

 

I've read the docs, but my naiviety made them a bit opaque as to this particular scenario. 

 

Thank you!

1 ACCEPTED SOLUTION

SeanHatfield
2,830 Views

For NAS lifs, the interface will failover to another port within the broadcast domain, even if that means moving it to some other node in the cluster.  The data lif and the aggregate don't have to be on the same node.  Ensure that the broadcast domain(s) contain ports from both nodes.

 

 

For SAN lifs, the interface does not fail over.  Instead SAN lifs are placed on both nodes, and host side multi-pathing and ALUA take care of path selection.

 

In either case a storage failover is not needed.  

If this post resolved your issue, help others by selecting ACCEPT AS SOLUTION or adding a KUDO.

View solution in original post

2 REPLIES 2

SeanHatfield
2,831 Views

For NAS lifs, the interface will failover to another port within the broadcast domain, even if that means moving it to some other node in the cluster.  The data lif and the aggregate don't have to be on the same node.  Ensure that the broadcast domain(s) contain ports from both nodes.

 

 

For SAN lifs, the interface does not fail over.  Instead SAN lifs are placed on both nodes, and host side multi-pathing and ALUA take care of path selection.

 

In either case a storage failover is not needed.  

If this post resolved your issue, help others by selecting ACCEPT AS SOLUTION or adding a KUDO.

nhwanderer
2,816 Views

Thanks you, that's very helpful. We'll be running NAS (nfs to vmware).

Public