Unsolved
6 Posts
0
5651
Dell S4148 VLT
So I have a rather large deployment using a couple S4148F's which are connected via VLT. In the configuration guide for OS10, is shows very basic configuration steps to setup VLT. Create your domain, VLTi, backup, etc. And then when it comes to port channels it shows Switch A Port Channel 20 w/VLT Port Channel 10 and then Switch B Port Channel 30 w/VLT Port Channel 10. I am uplinking to a stack of N2048s via LACP. In a more advanced configuration using OSPF in the same guide for VLT setup, they use all the same ID for the port channel and VLT port channel for both switches.
The N2048 should be configure with a port channel matching the VLT port channel correct?
Does the Port Channel in each switch have to differ from one another and the only thing matching is the VLT Port Channel? Can they all be the same. If not why?
Should this example work?
Switch A
interface vlan10 description "Data LAN" no shutdown ip address 10.10.10.2/24 ! vrrp-group 10 priority 200 virtual-address 10.10.10.1 ! interface port-channel10 description "UPLINK to N-Series Stack A Port Te1/0/1" no shutdown switchport mode trunk switchport access vlan 1 switchport trunk allowed vlan 10 vlt-port-channel 10 ! interface ethernet1/1/1 no shutdown channel-group 10 mode active no switchport flowcontrol receive on flowcontrol transmit on !
Switch B
interface vlan10 description "Data LAN" no shutdown ip address 10.10.10.3/24 ! vrrp-group 10 priority 200 virtual-address 10.10.10.1 ! interface port-channel10 description "UPLINK to N-Series Stack A Port Te4/0/2" no shutdown switchport mode trunk switchport access vlan 1 switchport trunk allowed vlan 10 vlt-port-channel 10 ! interface ethernet1/1/1 no shutdown channel-group 10 mode active no switchport flowcontrol receive on flowcontrol transmit on !
I have some intervlan routing issues and I am curious if this may be the cause. Although I really need to know why because things seem to work OK and then occasionally does notwhen doing things like adding a new device on a particular stack that is part of a new VLAN to that stack.
bealdrid2
117 Posts
0
September 26th, 2018 05:00
The port channel on each VLT peer (in this case, your S4148's) does not need to be different. I don't really see an issue with the VLT part of your config. The "vlt-port-channel" is more or less just an "index", where the port channels on each peer that connect to the same upstream device would have the same "vlt-port-channel" id. I really don't think it matters what the id is, as long as its consistent. It doesn't have to match the numbering of the port channel. That said, I don't think it would hurt for them to be the same.
Obviously since the N series switch/stack doesn't "talk" VLT you would just do a standard LACP port channel on that. Can you post the config for that?
If you go in to a little more detail regarding your problem I may be able to assist more. You mention inter-vlan routing issues. Could you provide a couple examples? Also I see you have VRRP set up. Have you considered simply using the "peer-routing" functionality and ditching VRRP altogether? Works great for us and a lot simpler.
RyloAT
6 Posts
0
November 21st, 2018 17:00
The engineer likes the idea of having a virtual IP address, which is why we are using VRRP. I do have peer routing enabled as well. I saw in other setups where running both is fine, just uses more resources. I have a basic configuration on the N series stack in regards to the uplink. Te1/0/1, Te4/0/2 both just channel group 10 and the port channel 10 config is just switchport mode trunk.
Problem I run into is something things seem to run into a situation where I am missing packets. from a pc connected to the access stack (n series) i can run a constant ping to either vlt member and the virtual IP. The primary member and the virtual I get a constant ping return, but the backup vlt member I get a ping then 2 timeouts and this is consistant as long as the constant ping is rolling. While this occurs, I will get half the web management interface for the firewall, but this may be something in the firewall as well because standard web traffic is fine. a refresh brings my web interface up just fine. The firewall is actually 2 in a high availability setup each with a lag into each vlt member.
My original concern is the basic ping communication from the stack to the vlt members and can troubleshoot the other further. I had more than a thousand users on the system with no reported issues.
Another concern in the vlt setup is I am running rapid-pvst on the vlt members. While the OS10 manual said this is supported. When I do a show run, it hangs for a few seconds before the spanning-tree vlan # priority statement. I of course made my primary vlt member root for all the vlans and the secondary the next in line for rapid-pvst. That pause in the show run is just concerning.