Hey all,
I am curious about a setup and wanted to see if anyone could discover any fallacies that could occur with this method. I have performed this on my home lab and it seems to have been successful. Just wanted to share my insight and see what others thought.
I want to remotely enable LACP bundle between local switch and far end switch without losing network connectivity. Currently both switches have redundant links to one another and spanning-tree is set to rapid-pvst. One link is being blocked on the far end switch as expected. Local switch is designated as the root switch. I will designate local switch as Switch A, and remote switch as Switch B.
My theory and so far I have been able to do so is the following:
1.) Shut down the blocked interface on Switch B
2.) Configure trunking paramters including native vlan and encapsulation on Switch B
3.) Join the shut interface to a new channel-group using mode active on Switch B
3.) Configure Switch A interface trunking parameters on interface that connects to the shut interface on SwitchB
4.) Join the Switch A interface to a new channel-group using mode active.
5.) No shut the interface on Switch B and allow the link to come up. ( I anticipate STP to block this link once again since it is redundant)
6.) Shut down the primary interface on Switch B (I anticipate STP to move over to the port that was previously being blocked as a result. I expect to lose a brief instance of connectivity before link fails over).
7.) Configure trunking parameters on primary interface on Switch B and join it to the same channel-group as the previous
8.) Configure trunking parameters on Switch A interface and join it to the local channel-group.
9.) No shut the remote Switch A interface.
From this sequence, I expect to lose minimal connectivity with the only instance being Step 6 where STP fails over to the previously blocked link which is now in LACP channel group. Once Step 9 is complete, I expect the bundle on both ends to sync up and the port-channel will be created without losing complete connectivity to remote switch.
Does this seem like a reasonable approach? Let's pretend that Switch A and B are far apart from one another and having to walk from switch to switch can be very time consuming if local console is needed.