OCI 6.4.0 introduces a new device type, "NPV Switch".
Cisco NPV mode / Brocade Access Gateway mode are typically used (often optionally) on edge devices - smaller rack switches, or switch like devices within blade server architecture. They do not participate in FC zoning, but are subject to it. They logically act like a passthrough device to the upsteam fabric - allowing the physically attached hosts to make NPIV logins to the upstream fabric.
OCI 6.4.0 will cease reporting NPIV attached ports as fc1/2-a , and instead, a "hub switch" will be shown in the NPV switch view to represent the devices on that port.
OCI 6.4.0 out of the box will support discovery of Cisco NPV mode devices. HOWEVER, this will not happen automagically - your fabric aware datasource will NOT discover these devices. You will need to create new Cisco datasources, one per NPV device, to discover these devices. These methodology to control the risk of chaos - some customers may have hundreds of these devices on a fabric, so having a fabric aware datasource automatically discover them might make acquisition time for that datasource exceedingly slow.
Note - OCI has a cli method to script the creation of datasources, so if you have > 20 NPV devices you want to discover, pursuing this may be a good idea.
OCI 6.4.0 does not have Brocade Access Gateway support out of the box. However, we are starting a public beta for it - basically, you will be able to deploy a Brocade CLI datasource patch that *will* allow your existing Brocade CLI datasource to discover your AG devices through fabric discovery
Brocade AG and Cisco NPV devices will both appear in the NPV switches view - when "actively" discovered by a datasource, we will know the make/model/firmware of the AG/NPV mode device, and discover all of its ports. In the future, we are looking to support performance collection for these devices' individual ports.
Finally, having the model of NPV mode device in OCI may allow us to create new datasources in the future, for other vendor's edge fabric devices that are not truly FC switches (which in the eyes of OCI, need to participate in zoning, have a domain ID, etc)
Thanks for the information. I have been waiting since long time for this NPV switches support in OCI. I already downloaded 6.4 version, will upgrade my servers this week.
I have more than 50 switches configured in NPV mode, so I would like to know more on discovering them using CLI method. I checked the install and admin guide but nothing is there about discovering NPV.
Interesting. I'm seeing NPV devices show up as "hubs" in the topologies without any action on the customer's part.
Something interesting is also happening with customers that upgrade to 6.4. The "hub" adds one to the hop count from host to array. If that violates the standing global FC policy, it generates a SAN violation message. Changing the global policy causes those violations (and the hosts that all suddenly show up with violations in the Hosts Table), to automatically disappear.
Yes, I found the CLI method to configure NPV switches. It is not any rocket science, you have to create an excel file which contain all the parameters that you manually enter while adding a data source. Then you have to import that excel file in OCI. You will not find about it in any document of OCI, I don't know why it is not published in yet. Anyway let me know if you need further help in creating that excel and how to import it.
I also faced the same error message "has no domain manager". You have to enter the SNMP community string to resolve it.
The "no domain manager" is a known, solved issue that we have a patch for. You are likely to encounter this issue when attempting to discover MDS9[1-2]xx running in NPV mode - it does not occur on Nexus devices in NPV mode.
I believe this issue will be resolved in 6.4.1 and data source service pack 220.127.116.11. I know 6.4.1 is anticipated for the very end of next month.