01-13-2014 09:42 AM - edited 03-07-2019 05:32 PM
Hi everybody,
I can´t answer my question, why a part of my network behaves like it does. I´m pointing to the scenario that you can see on the image attached.
I´ve also tried to find the correct document, that explains this behaviour, but i didn´t find it.
The whole network is configured with PVST. Core1 is root bridge, core2 is backup. the question is, why switch-a and switch-b have their portchannels in blocking mode and are using the slower 1Gb links as the root port.
Best Regards
Sebastian
Solved! Go to Solution.
01-13-2014 11:03 AM
Sebastian
The root bridge, core1 is advertising out BPDUs with a cost of 0 because it is the root bridge. Switches A and B have direct connections to the core 1 so the the cost of the interface to get to the root is 4 via it's direct 1Gbps connection. They also receive BPDUs from the root bridge via core 2 but the cost here is -
cost from core 2 to core 1 = 3 + the cost to get from switch A/B to core 2 = 3 = 6
So the lowest cost path to root is via the direct connections. If core 2 was the root bridge then you would find that the 1Gbps links are blocked and the port channels to core 2 would be forwarded.
If you wanted to make the port channels the forwarding links in your current setup then you can change the port cost on the 1Gbps links on switches A and B by using the "spanning-tree cost
So you would need to make the cost > 6 for it to favour the port channels. If you do this be aware there will be an outage while STP reconverges.
Jon
01-13-2014 11:02 AM
Hi,
If your diagram is correct, switch A and B are connected to both Core-1 and 2 and since core-1 is the root that is the shortest path to the root (cost 4) from both switch A and B.
HTH
01-13-2014 11:03 AM
Sebastian
The root bridge, core1 is advertising out BPDUs with a cost of 0 because it is the root bridge. Switches A and B have direct connections to the core 1 so the the cost of the interface to get to the root is 4 via it's direct 1Gbps connection. They also receive BPDUs from the root bridge via core 2 but the cost here is -
cost from core 2 to core 1 = 3 + the cost to get from switch A/B to core 2 = 3 = 6
So the lowest cost path to root is via the direct connections. If core 2 was the root bridge then you would find that the 1Gbps links are blocked and the port channels to core 2 would be forwarded.
If you wanted to make the port channels the forwarding links in your current setup then you can change the port cost on the 1Gbps links on switches A and B by using the "spanning-tree cost
So you would need to make the cost > 6 for it to favour the port channels. If you do this be aware there will be an outage while STP reconverges.
Jon
01-13-2014 11:13 AM
Because the cost to the root bridge via core-2 is 6 via the Portchannels (3+3). The cost via the Gigalinks is 4. You could tweak the cost on the Portchannel if you want to have that as root port. If I remember correctly the cost won't be lower than 3 until you reach 10GE.
So the solution would be to manually enter cost if you want the topology to follow the faster links.
Daniel Dib
CCIE #37149
Please rate helpful posts.
01-14-2014 02:53 AM
Hey guys,
thanks for the explanation, that answered my question! I´ve just looked for the IOS Command, that shows me the total root costs. Here the command, if somebody didn´t know it:
sh span vlan [vlan-id] root
Have a nice day,
Sebastian
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide