I have a fairly simple configuration - two Nexus 3048 ( A and B ), both enterprise licensed. One 10G port on switch A is the start of a fibre ring around a number of HP switches, the ring is closed on the same port on switch B. One 10G port is to route to a different building - this one isn't my problem ( yet ) (the remaining 10G pair are my peer links).
I have the vPC peer links and keep alive configured and working fine. I've configured a number of ways using either VRRP or HSRP. In both cases I can ping the VLAN IP address's and the 'virtual' IP from within the switches themselves. I cannot ping or reach the IP's from outside the switch regardless of the port mode (trunk or access), all vlans allowed or specific.
Ultimately I want eth1/1 on each switch to form an LACP pair with a HyperV host array but testing this is a pain in the butt without the host since the Cisco's refuse to bring the ports up unless LACP is seen - which most laptops can't do. I did have this 'sort of' functional but then started to diagnose a DHCP issue - discover received, offer made but that's where the exchange ended. The DHCP discover broadcasts came in, the offer was made but the request and ack were never triggered but the offer was present at the device.
I am not so familiar with these newer techniques on the Nexus boxes - not touched a switch in years. Can somebody have a scan of the config below, also can somebody offer a source of knowledge that is up to date that goes into how to pull all this together because much of what I've found seems to be out of date. The Cisco docs are OK if you're digging into individual topics but pulling everything together isn't a document that I have found yet. Everything I've read says this should work - but it doesn't - the ring of switches aren't cisco (they're Aruba 2930's with an SFP+ module) there is broadcast traffic appearing, the Aruba's also know the 3K's are there (they show the switch name as the neighbour). I'm trying to expand the knowledge whilst the opportunity presents.
My server dhcp / dns server is currently on eth1/1 in the config below (this will be one side of the hyperV).
I do have peer-gateway and also layer3 peer-router set on the vpc domain. The configs below leave out the peer link stuff and are HSRP but I've tried VRRP too.
I have port states showing no operational member which isn't true, cables are connected appropriately. eth1/50 shows connected to vlan 9 but the associated port channel (po9) shows noOperMem. eth1/1 shows connected 'trunk' but po10 shows noOperMem .... I can get them 'connected' if I don't use active mode and use the default of 'on' but it makes no difference.
NEXUS-3048-A config
vl 9
name device-net
int vl 9
no shut
description **HSRP**
ip address 10.9.1.252/24
hsrp 9
preempt delay minimum 300
priority 130
ip 10.9.1.254
int eth1/50
description **Aruba Link**
channel-group 9 mode active
no shut
int port-channel 9
description **Aruba Circuit**
switchport mode trunk
no shut
int eth1/1
channel-group 10 mode active
int port-channel 10
switchport mode trunk # I have tried this as 'Access' / 'Trunk', specific VLAN, all VLAN
no shut
router eigrp 1
log-adjacency-changes
int eth1/52
description **Routed Link A**
channel-group 1003 mode active
no shut
vl 1003
int port-channel 1003
no switchport
ip router eigrp 1
NEXUS-3048-B config
vl 9
name device-net
int vl 9
no shut
description **HSRP**
ip address 10.9.1.253/24
hsrp 9
preempt delay minimum 300
priority 110
ip 10.9.1.254
int eth1/50
description **Aruba Link**
channel-group 9 mode active
no shut
int port-channel 9
description **Aruba Circuit**
switchport mode trunk
no shut
int eth1/1
channel-group 10 mode active
int port-channel 10
switchport mode trunk
no shut
router eigrp 1
log-adjacency-changes
int eth1/52
description **Routed Link B**
channel-group 1003 mode active
no shut
vl 1003
int port-channel 1003
no switchport
ip router eigrp 1
Being able to learn something is not the same thing as being able to do it for real. The only thing that exams prove is your memory.
