05-11-2019 11:56 AM - edited 05-11-2019 12:06 PM
Hello, normally our shop uses two physical core(Core1 and Core2) switches in a HSRP configuration using PAgP between the two switches. Each of these switches has a redundant layer 3 link to devices further upstream in the network. If Core A fails, Core B can still route traffic upstream and vise versa. So far so good, everyone is happy.
Now they want me to do the same thing except they want it done using a stackwise configuration for both of these core switches. So Core1 and Core2 are in one logical switch configuration.
Knowing this, how do I keep my redundant links to the upstream devices if this is just one logical switch? I must be having a mind melt down, because I can't get my head around this.
Any help would be greatly appreciated.
Solved! Go to Solution.
05-11-2019 02:21 PM
In a logical StackWise configuration, I can't have two different default routes going to two different locations.
There is no way, as far as I know, to have member 1 switch with a default route and switch member 2 to have a different default route.
Once you stack the 2 3850s together, they logically become one switch. You only manage one switch, configure one switch (the master). So, think of one switch. Now, you can have 2 default routes to 2 upstream devices. So, say on the 3850 stack, interface 1/0/1 is connecting to one upstream device and for redundancy interface 2/0/1 is connecting to the other upstream device and you want 1/0/1 to be the primary. So, you configure a default route and point it to the next hop IP which is the primary upstream device. Now, you configure another default route with higher AD pointing to the other upstream device.
example:
ip router 0.0.0.0 0.0.0.0 192.168.1.2 (this is for the primary upstream device)
ip router 0.0.0.0 0.0.0.0 192.168.2.2 10 (this is for the backup upstream device)
HTH
05-11-2019 12:21 PM
Its is the same way you doing now. onlything is you do not HSRP.
Switch 1 act a MAster and Switch2 Act as Slave, But both can work in Data path.
If switch 1 Fails Switch 2 act as master switch ( based on your configuration)
make sense ?
05-11-2019 01:01 PM
Not quite. In my original configuration with two physical switches, I have one layer 3 exit interface on each switch, each going to to a different address.
In a logical StackWise configuration, I can't have two different default routes going to two different locations.
There is no way, as far as I know, to have member 1 switch with a default route and switch member 2 to have a different default route.
05-11-2019 01:15 PM
I have one layer 3 exit interface on each switch, each going to to a different address.
how is your failover config for now for your ISP or uplink from your Local LAN ?
suggest to post some config of both the switches or HLD diagram ?
05-11-2019 02:18 PM
Currently Core 1 and Core 2 use HSRP between them.
Core 1 has a layer 3 interface to a upstream ISP
Core 2 has a layer 3 interface to a upstream ISP
These are two different routes.
I don't see a way to do this if these switches are in a single logical switch.
05-11-2019 02:21 PM
In a logical StackWise configuration, I can't have two different default routes going to two different locations.
There is no way, as far as I know, to have member 1 switch with a default route and switch member 2 to have a different default route.
Once you stack the 2 3850s together, they logically become one switch. You only manage one switch, configure one switch (the master). So, think of one switch. Now, you can have 2 default routes to 2 upstream devices. So, say on the 3850 stack, interface 1/0/1 is connecting to one upstream device and for redundancy interface 2/0/1 is connecting to the other upstream device and you want 1/0/1 to be the primary. So, you configure a default route and point it to the next hop IP which is the primary upstream device. Now, you configure another default route with higher AD pointing to the other upstream device.
example:
ip router 0.0.0.0 0.0.0.0 192.168.1.2 (this is for the primary upstream device)
ip router 0.0.0.0 0.0.0.0 192.168.2.2 10 (this is for the backup upstream device)
HTH
05-11-2019 02:40 PM
I believe you are partially correct, having a floating static route will work, but I will also need to using some sort of interface tracking. Again, I'm not familiar with stackwise setup.
05-11-2019 03:27 PM
Stacking is just a physical topology where you stack 2 or more switches together so, they can look like one. As I said, once it is configured, you are looking at one switch (logically). You only configure the master and the config gets pushed to the rest.
HTH
05-11-2019 11:40 PM
"I believe you are partially correct, having a floating static route will work, but I will also need to using some sort of interface tracking. Again, I'm not familiar with stackwise setup."
BB- I have asked one of the post, how are you managing now your failover?
05-12-2019 07:18 AM
As I said I have two physical switches which are running HSRP between them. Each switch has a separate physical connection to a upstream device.
05-13-2019 11:32 PM
05-13-2019 11:42 PM
Hi
Reza has explained it well, just to try to explain further. It is still 2 physical boxes, however they are treated as if they where one box. So if you configure 1/0/1 as a routed interface with 192.168.1.1 as the example from Reza, and 2/0/1 with 192.168.2.1, you also configure the routes as Reza mentioned. Then if all is good both interfaces are alive and the route to 192.168.1.2 is then one in routingtable.
If then switch one dies, then interface 1/0/1 is not available anymore so the routing will put in the the route to 192.168.2.2 instead.
Then when switch one is back in business, and the 1/0/1 is up again, the original route will be inserted.
/Mikael
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