Unsolved
This post is more than 5 years old
3 Posts
0
31393
June 19th, 2010 11:00
High latency and dropped packets managing 6224 stack with routing enabled.
I have a 6224 and 6224F configured in a stack. When I turn routing on, I get high latencies and dropped packets that make manging the switch through the routing interface impossible. Before I enabled routing I created an OOB ip address on an unused VLAN and I made that the management VLAN. I then turned on routing for my VLANs (2 and 5 in this example) but when I try to manage in those ip addresses, I drop so many packets I can't reliable load the web interface.
As soon as I turn off routing an re-assign the mangement vlan to 5, I'm able to configure again.
Any ideas of what can be causing that?
exit
stack
member 1 1
member 2 3
exit
ip address 10.255.255.10 255.255.255.0
ip default-gateway 10.255.255.1
ip address vlan 4093
ip routing
router rip
no enable
default-metric 1
exit
interface vlan 2
routing
ip address 172.20.23.254
exit
interface vlan 5
routing
ip address 172.20.3.1
exit
interface vlan 99
exit
spanning-tree bpdu filtering
spanning-tree mode stp
bridge multicast filtering
ip igmp snooping
router ospf
no enable
exit
!
0 events found
bh1633
909 Posts
0
June 21st, 2010 14:00
When you enable routing on a vlan, you put the switch CPU in the vlan. This means broadcast and multicast traffic in the vlans will hit the switch CPU. You should try adjust the broadcast and multicast storm control on the interfaces. I suggest starting at the most restrictive setting. Here is an example:
interface ethernet 1/g1
storm-control broadcast level 1
storm-control multicast level 1
exit
If this does not work, post the COMPLETE output of the following when you are experiencing the issue. You may need to use the serial interface.
-"show running-config"
- "show process cpu"
- "show arp"
JustinY
1 Rookie
•
117 Posts
0
June 30th, 2010 23:00
I saw similar symptoms on my 6200s when they were receiving large amounts of undeliverable packets caused by a down syslog server. Each packet sent to the syslog server would cause the 6200s to create an icmp undeliverable message. The large quantity of them caused the 6200 CPU to spike and become unresponsive to any form of management. It was still able to route traffic but suffered from very high latency like you describe.
You can stop the 6200 from sending the icmp unreachable messages when it is unable to deliver a packet. This might prevent the 6200 CPU from being overwhelmed with sending icmp unreachables. You have to disable ip unreachables on each interface.
If this is the case for you then look into why there are so many undeliverable packets on your network.
Tempbrennanm
3 Posts
0
July 1st, 2010 06:00
Thanks very much for the suggestions, both of you! I will try this out at my next maintenance window and post my results.