Wired Intelligent Edge

After designating the secondary-member switch, is it necessary to cable the mgmt ports of the primary and secondary together and add the vsf split-detec mgmt-interface mgmt-interface command if the vsf devices are already stacked in a ring topology? Shouldn't the secondary already know to take over because of the stacking cables being in a closed loop (ring)?

How VSF members - no matter the topology used to form the VSF, Chain or Ring - can recognize the status of the stack fragments (in case of a split) and manage a potential Split Brain scenario if there isn't a 3rd party mechanism arbitrating the VSF members status from another PoV (and the interaction among VSF members' OoBM ports is going to act as that external mechanism with regard to Split Brain detection)?

If you don't want to use OoBM you need a MAD (Multi-Active Detection) 3rd party device connected to all VSF members by using Ethernet interfaces.

------------------------------
Davide Poletto
------------------------------

Original Message:
Sent: Apr 28, 2022 09:53 AM
From: Unknown User
Subject: vsf split-detec mgmt-interface mgmt-interface

After designating the secondary-member switch, is it necessary to cable the mgmt ports of the primary and secondary together and add the vsf split-detec mgmt-interface mgmt-interface command if the vsf devices are already stacked in a ring topology? Shouldn't the secondary already know to take over because of the stacking cables being in a closed loop (ring)?

Split-detection is used in the event of a failure of either VSF links or members in a chain topology that results in the primary and secondary being on opposite sides of the split; the secondary would detect the loss of the primary and failover to the Conductor role, but if the primary is still up and running in the other fragment and split-detection were not configured and working, you would end up with two stacks with the same MAC address and IP addresses, which can cause severe network disruption if there are uplinks/downlinks that remain active on both fragments.

Split-detection prevents this scenario by using the management port connections as a backup method of determining whether the primary is still powered on and operating after the normal VSF link connection from the secondary is lost; if the primary is determined to still be running after a split occurs, the fragment containing the secondary shuts down all ports except VSF link ports and the management port. When the failed links/members recover and connectivity to the primary is re-established, the 'inactive' secondary fragment automatically reboots to rejoin the stack.

------------------------------
Matt Fern
Sr. Technical Marketing Engineer, Aruba Switching
Aruba, a Hewlett Packard Enterprise company
------------------------------

Original Message:
Sent: Apr 28, 2022 09:53 AM
From: Unknown User
Subject: vsf split-detec mgmt-interface mgmt-interface

After designating the secondary-member switch, is it necessary to cable the mgmt ports of the primary and secondary together and add the vsf split-detec mgmt-interface mgmt-interface command if the vsf devices are already stacked in a ring topology? Shouldn't the secondary already know to take over because of the stacking cables being in a closed loop (ring)?