Print | Rate this content

HP ProCurve 8206zl Switch Base System - Network Utilization Goes to 100 % in Random Cycles

Issue

3 HP ProCurve E8206 ZL switches were installed and configured for (2)10 Gig UpLinks between Switch 1 and Switch 2 and between Switch 1 and switch 3. Then between Switch 2 and Switch 3 a 1 Gig Uplink was configured. A different Vlan number was assigned to each link. Spanning-Tree was not been enabled and spanning-tree bpdu-filters were applied to the uplink ports to block BPDUs. The network utilization spikes randomly up to 100% sometimes for hours, sometimes for minutes and then it comes back down. The traffic is made up of mostly Broadcast and Multicast traffic. Since Routing has been enabled and each Uplink between Switches is in a different vlan, the end-user said no loops exist.

Solution

The configuration for the ProCurve E8206ZL Switches and a network map were emailed in by the customer. This Network map is in Figure 1.

Figure 1: Network Map

In each of the ProCurve E8206ZL Switches the ports for each uplink were tagged in the respective vlans as shown below in configs.

SW 1

vlan 1
name "DEFAULT_VLAN"
untagged A1-A4,B1-B24,C1-C4,D1-D4,E1-E4,F9-F24
ip address dhcp-bootp
no untagged Trk1
exit
vlan 100
name "10G VLAN to SW3"
ip address 10.101.46.1 255.255.255.252
tagged D1
exit
vlan 300
name "Connection to SW2"
ip address 10.1.136.1 255.255.255.252
tagged D4
exit

SW 2

vlan 1
name "DEFAULT_VLAN"
untagged A1-A4,B1-B24,C1-C4,D1-D4,E1-E4,F9-F24
ip address dhcp-bootp
no untagged Trk1
exit
vlan 200
name "1G VLAN to SW3"
ip address 10.101.46.5 255.255.255.252
tagged B1
exit
vlan 300
name "Connection to SW1"
ip address 10.1.136.2 255.255.255.252
tagged D4
exit

SW 3

vlan 1
name "DEFAULT_VLAN"
untagged A1-A4,B2,B21-B24
ip address dhcp-bootp
no untagged B1,B3-B20,G1-G24,I1-I24,J1-J24,K1-K24,L1-L24
exit
vlan 100
name "10G VLAN to SW1"
ip address 10.101.46.2 255.255.255.252
tagged A1
exit
vlan 200
name "1G VLAN to SW2"
ip address 10.101.46.6 255.255.255.252
tagged B2
exit

In looking at Vlan 1, the Default vlan, the uplink ports were also untagged in this vlan creating a layer 2 loop. Without spanning-tree running this caused Broadcast and Multicast traffic to loop around the network. To make matters worse Vlan 1 got its IP address via the IP address dhcp-bootp command under vlan 1 which sends a broadcast out to get its IP address.

Solution:

The uplink ports D1,D4 in SW1, B1,D4 in SW2 and A1,B2 in SW3 were all removed from the Vlan 1 Default vlan in each switch via the No untagged Port command placed in each Vlan config as shown below in the configuration for Vlan 1 for each switch, removing the loop.

SW 1

vlan 1
name "DEFAULT_VLAN"
untagged A1-A4,B1-B24,C1-C4,D2-D3,E1-E4,F9-F24
ip address dhcp-bootp
no untagged D1,D4,Trk1
exit

SW 2

vlan 1
name "DEFAULT_VLAN"
untagged A1-A4,B2-B24,C1-C4,D1-D3,E1-E4,F9-F24
ip address dhcp-bootp
no untagged B1,D4,Trk1
exit

SW 3

vlan 1
name "DEFAULT_VLAN"
untagged A2-A4,B21-B24
ip address dhcp-bootp
no untagged A1,B1-B20,G1-G24,I1-I24,J1-J24,K1-K24,L1-L24
exit

Provide feedback

Please rate the information on this page to help us improve our content. Thank you!