The eth-trunk using lacp keeps on flapping because of lots of STP packets received.

Publication Date:  2013-03-31 Views:  476 Downloads:  0
Issue Description

Huawei switch was connecting to H3C switch with LACP static mode by eth-trunk4, the eth-trunk 4 status keeps on flapping every few minutes.
Alarm Information

2/13/2013 4:24:22 PM        x.x.249.5 Notice x-x-249.5 The status of the trunk member went Down. (TrunkName=Eth-Trunk4, PortName=GigabitEthernet0/0/3)
2/13/2013 4:22:27 PM        x.x.249.5 Notice x-x-249.5 The status of the trunk member went Up. (TrunkName=Eth-Trunk4, PortName=GigabitEthernet0/0/1)
2/13/2013 4:22:14 PM        x.x.249.5 Notice x-x-249.5 The status of the trunk member went Down. (TrunkName=Eth-Trunk4, PortName=GigabitEthernet0/0/1)
2/13/2013 4:21:21 PM        x.x.249.5 Notice x-x-249.5 The status of the trunk member went Up. (TrunkName=Eth-Trunk4, PortName=GigabitEthernet0/0/2)
2/13/2013 4:21:13 PM        x.x.249.5 Notice x-x-249.5 The status of the trunk member went Down. (TrunkName=Eth-Trunk4, PortName=GigabitEthernet0/0/2)
2/13/2013 4:21:07 PM        x.x.249.5 Notice x-x-249.5 The status of the trunk member went Up. (TrunkName=Eth-Trunk4, PortName=GigabitEthernet0/0/4)
Handling Process
1.Use port mirror to capturing packets transfering along the eth-trunk, make sure that the LACP packets are sent normally by H3C switch and received by S5700HI. The result is OK.
2.Check the CPU queue of LACP status, when the issue occured there's lots of packets received (108 per second).Which is not normal.
BCM.0> show c cpu
PERQ_PKT(0).cpu0            108/s
3.Next step, check what the packets is under this queue,information about packets sent to the CPU showed that many STP protocol packets were sent to the CPU (more than 100 packets per second).
[CHRM-TRNK-PARIMALAHOMES-248.100-hidecmd]disp gfpi catch  slot  0 stat-pkt-info  receive  dstmac
RECEIVE DST-MAC : 0x00001540 --- 0180-c200-0000 // 0x1540 is the number of packets, and 0180-c200-0000 indicates the protocol packets with destination MAC address 0180-c200-0000 (STP packets).
4.We established the same networking environment in our lab and sent STP protocol packets at a rate of 100 pps. LACP flapping occurred.
5.S5700HI provides commands to change priorities of protocol packets. Use the following commands to set a high priority for LACP packets (higher than priority of STP packets), so that other protocol packets will not cause LACP flapping in a complicated networking environment.
#
cpu-defend policy lacp
queue packet-type lacp 7
#
cpu-defend-policy lacp global
After changing the priority, the issue was resolved.
Root Cause
The eth-trunk worked normally in the last few week, with LACP static mode for interconnecting. Eth-trunk status keeping on flapping usually is because of the LACP protocol timeout, so there might be some reasons that make the LACP protocol work abnormal.
Suggestions
Usually there may be some STP packets in network, if there's some reason like loop, should be lots of STP packets, which will make other protocol work abnormally if with same priority. So in later versions of S5700, the default priority of LACP packets is higher than the default priority of STP packets.

END