#Broadcast domain question - multi-domain-reachability

1 messages · Page 1 of 1 (latest)

native sage
#

After moving to 9.12 this has started getting flagged for our clusters. Basically I have one subnet in a single vlan and an internal network that is used for client facing lifs and also for management. The Default BD was used when the management lifs were configured, and another BD was added when I setup the client facing lifs. These are all in the same network, but are in different broadcast domains as noted in the error. In reading it looks like I can use the merge command, but its not clear to me if it matters from which BD I merge the other into, its also not clear if there will be any disruption to traffic when that happens, and also not clear if it would try to rearrange failover-groups if the BD are merged. All of my clusters are configured this way, and have been since Netapp installed them some time ago. The error has only started getting logged after moving to the 9.12 firmware. I do have a ticket open on this, but its been somewhat frustrating and was hoping someone with a bit more networking knowledge might have a suggestion. There are currently no noticeable issues with any traffic, but the events have me concerned I should merge the BD.

The lifs do all show "multi-domain-reachability" when using the reachablity command.

ERROR vifmgr.lif.subnetMisconfig: LIFs in subnet 1.2.3.0/24 of IPspace "Default" are configured on ports in multiple broadcast domains: Default, another-bd

limpid pilot
#

Short of making a separate vlan to separate management and data, this will exist.

#

You could lop all the ports in the same BD m you run the risk of ONTAP inadvertently selecting an e0M port for data. It’s rare, but can and does happen

#

There really should be different vlans for different objectives.
Management
Customer facing data
Private NFS for VMware (no gateway)
Private iscsi for anything (no gateway)

versed warren
#

you can put all ports in one broadcast domain and then create a failover group without the e0M ports... but it's probably not worth the hassle, as long as ONTAP has enough ports that are fast enough, it will not select the e0M

limpid pilot
#

Unless the logic has truly changed, that’s not always the case. I’ve had customers with e0M mixed with data ports and ONTAP has in fact selected them essentially preventing data services. It’s rare but it does happen. The failover group may be the only way but how long before that is deprecated?