Physical Network Topology
Figure 1-1 Two direct connected devices ping time delay
Fault Description
When SwitchA pings SwitchB, the ICMP delay is very large. The specific information is as follows:
[CE12800_A]
ping 111.111.1.3
PING 111.111.1.3: 56 data bytes, press CTRL_C to break
Reply from 111.111.1.3: bytes=56 Sequence=1 ttl=255 time=52
ms
Reply from 111.111.1.3: bytes=56 Sequence=2 ttl=255 time=52
ms
Reply from 111.111.1.3: bytes=56 Sequence=3 ttl=255 time=51
ms
Reply from 111.111.1.3: bytes=56 Sequence=4 ttl=255 time=51
ms
Reply from 111.111.1.3: bytes=56 Sequence=5 ttl=255 time=51
ms
When pinging with the -r parameter, the delay is very small, as follows:
[CE12800_A]
ping -r 111.111.1.3
PING 111.111.1.3: 56 data bytes, press CTRL_C to break
Reply from 111.111.1.3: bytes=56 Sequence=1 ttl=255 time=22
ms
Record Route:
111.111.1.3
111.111.1.1
Reply from 111.111.1.3: bytes=56 Sequence=2 ttl=255 time=2
ms
Record Route:
111.111.1.3
111.111.1.1
Reply from 111.111.1.3: bytes=56 Sequence=3 ttl=255 time=2
ms
Record Route:
111.111.1.3
111.111.1.1
Reply from 111.111.1.3: bytes=56 Sequence=4 ttl=255 time=2
ms
Record Route:
111.111.1.3
111.111.1.1
Reply from 111.111.1.3: bytes=56 Sequence=5 ttl=255 time=2
ms
Record Route:
111.111.1.3
Configuration Files
l Configuration on SwitchA
#
interface Vlanif1111
ip address 111.111.1.1 255.255.255.0
interface 10GE1/0/1
port link-type trunk
port trunk allow-pass vlan 5 10 to 30 300 1111 3000
stp disable
device transceiver 10GBASE-COPPER
#
l Configuration on SwitchB
#
interface 10GE2/0/1
port link-type trunk
port trunk allow-pass vlan 10 to 30 1000 1111
stp disable
device transceiver 10GBASE-COPPER
#
interface Vlanif1111
ip address 111.111.1.3 255.255.255.0
icmp name host-unreachable receive disable
icmp name redirect receive disable
#
Troubleshooting Procedure
Step 1 Check that SwitchB receives the ICMP packet normally and finds that the packet has been sent from the chip to the CPU.
<CE12800_B>
display cpu-defend statistics packet-type icmp slot 1/4
Statistics(packets) on slot 1/4 :
--------------------------------------------------------------------------------
PacketType
Total Passed Total
Dropped Last Dropping Time
Last 5 Min Passed Last 5 Min Dropped
--------------------------------------------------------------------------------
icmp
5
0 -
5
0
Step 2 Check whether the CPU receives ICMP packets and speed limit on the CPU. The details are as follows.
<CE12800_B>
debugging ip icmp
Oct 21 2018 21:46:48.225 switch-C %PP4/7/icmp_packet(d):CID=0x8066274d;ICMP
Receive: echo(Type=8, Code=0, Chksum=0xd3fe),IfIndex = 253, Vrf = 0, EventId =
0, Flag = 0x200, Src = 111.111.1.1, Dst = 111.111.1.3,ICMP Id = 0x2400, ICMP
Seq = 0x1
Oct 21 2018 21:46:48.233 switch-C %PP4/7/icmp_packet(d):CID=0x8066274d;ICMP
Send: echo-reply(Type=0, Code=0, Chksum=0xdbfe),IfIndex = 253, Vrf = 0, EventId
= 0, Flag = 0x100200, Src = 111.111.1.3, Dst = 111.111.1.1,ICMP Id = 0x2400,
ICMP Seq = 0x1
After ICMP debugging is enabled, it is found that no ICMP is sent to the CPU.
Step 3 Check whether the interface board fast returns ICMP packets.
<CE12800_B>
display icmp fast-reply statistics slot 1/4
------------------------ Display ICMP Statistics
-------------------------------
Received
packets:
request
packets:
5
invalid request packets: 0
failed to get
vrf:
5
destination is not host
ip:
0
failed to get ctrl
word:
0
Send
packets:
successful reply
packets:
5
failed reply
packets:
0
When the packet is ICMP fast return, the interface board processes the packets at the same time, and these data interface boards do not need to be processed.
Step 4 Turn off the ICMP echo-reply fast function.
[CE12800_B] icmp echo-reply fast disable
Step 5 Verify on SwitchA as follows.
<CE12800_A>
ping
111.111.1.3
PING 111.111.1.3: 56 data bytes, press CTRL_C to break
Reply from 111.111.1.3: bytes=56 Sequence=1 ttl=255 time=15
ms
Request time out
Reply from 111.111.1.3: bytes=56 Sequence=3 ttl=255 time=2
ms
Request time out
Reply from 111.111.1.3: bytes=56 Sequence=5 ttl=255 time=1
ms
--- 111.111.1.3 ping statistics ---
5 packet(s) transmitted
3 packet(s) received
40.00% packet loss
round-trip min/avg/max = 1/6/15 ms
The problem has been solved.
Root Cause
After the ICMP unreachable packet is not received funcion is enabled on the device, the packet is first processed by ICMP. After the packet is processed, the packet is sent to the CPU for processing. This causes excessive performance.
