No relevant resource is found in the selected language.

This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies. Read our privacy policy>

Reminder

To have a better experience, please upgrade your IE browser.

upgrade

Network Disconnecting Issue caused by Eth-Trunk mode change on E9000

Publication Date:  2018-01-30 Views:  105 Downloads:  0
Issue Description

1.         E9000 server comes across issue of intermittent disconnection of network.

2.         After adjusting the mode of the Eth-Trunk of switchboard to LACP dynamic, then the business network will be affected

The  figure shows the E9000 server network topology.

 

 

 

 

 

 

 

 

 

 

Handling Process

 Log Analysis of Blade IP Intermittent Disconnection Issue
Through the switchboard logs can be seen that frequently MAC address drift of the blade NIC between the switchboard internal port and switchboard uplink port.
===============================================================================
                 display arp track
===============================================================================
Operate Flags: M - Modify, D - Delete
--------------------------------------------------------------------------------
Op IP-Address      MAC-Address    VLAN Old-Port     New-Port     System-Time
--------------------------------------------------------------------------------
M  172.20.35.71    e472-e2ef-f180 42   Eth-Trunk1   10GE2/1/2    01-23 00:18:49
M  172.20.35.72    e472-e2ef-f4d0 42   Eth-Trunk1   10GE2/9/2    01-23 00:18:49
M  172.20.35.72    e472-e2ef-f4d0 42   10GE2/9/2    Eth-Trunk1   01-23 00:18:49
M  172.20.35.71    e472-e2ef-f180 42   10GE2/1/2    Eth-Trunk1   01-23 00:18:50
M  172.20.35.72    e472-e2ef-f4d0 42   Eth-Trunk1   10GE2/9/2    01-23 00:18:50
M  172.20.35.72    e472-e2ef-f4d0 42   10GE2/9/2    Eth-Trunk1   01-23 00:18:50
M  172.20.35.71    e472-e2ef-f180 42   Eth-Trunk1   10GE2/1/2    01-23 00:18:51
M  172.20.35.71    e472-e2ef-f180 42   10GE2/1/2    Eth-Trunk1   01-23 00:18:51
M  172.20.35.72    e472-e2ef-f4d0 42   Eth-Trunk1   10GE2/9/2    01-23 00:18:51
M  172.20.35.72    e472-e2ef-f4d0 42   10GE2/9/2    Eth-Trunk1   01-23 00:18:51
M  172.20.35.71    e472-e2ef-f180 42   Eth-Trunk1   10GE2/1/2    01-23 00:18:52
M  172.20.35.71    e472-e2ef-f180 42   10GE2/1/2    Eth-Trunk1   01-23 00:18:52
M  172.20.35.72    e472-e2ef-f4d0 42   Eth-Trunk1   10GE2/9/2    01-23 00:18:52
M  172.20.35.72    e472-e2ef-f4d0 42   10GE2/9/2    Eth-Trunk1   01-23 00:18:52

The MAC addresses that drift occurring include:472-e2ef-f180(10GE2/1/2)、e472-e2ef-f4d0(10GE2/9/2).
For example, a packet destined to 10GE2/1/2 (an internal port of switchboard) is sent to Eth-Trunk 1 when a MAC drift occurs. As a result, packets corresponding to 10GE2/1/2 cannot be obtained normally, resulting in packet loss situation, then blade IP intermittent disconnection.

 Cause of MAC Address Drift
1. Eth-Trunk1 on the E9000 switchboard is in manual load balancing mode
2. The ports of Cisco switch which connecting to E9000 are independent, they didn’t configure port channel.

In this situation, the broadcast packets from the blade NIC card are sent back to the Eth-Trunk 1 of the switch board through the Cisco core switch. As a result, the MAC addresses of blade NIC card are learned on both the internal ports of the switchboard and Eth-Trunk 1, causing MAC address drift.

The process of MAC address learning forwarding is shown in the figure below:

 

1 –Blade NIC sends broadcast packages to switchboard, the switchboard learns the NIC MAC in 10GE 2/1/2 port
2 –The switchboard forwards packets from 10GE 2/1/2 to Eth-Trunk 1
3 - The switchboard sends packets from Eth-Trunk 1 to the core switch
4 –The core switch is connected to the E9000 as an independent port, and packets are forwarded from the core switch
5 –The packets are sent from the core switch to Eth-Trunk 1 on the E9000 switchboard. The switchboard learns the MAC address of the NIC on Eth-Trunk 1

 

 

Log Analysis of Business Network Down Issue
When configure LACP dynamic on E9000 Eth-trunk1, while cisco doesn’t configure port group, LACP dynamic failed to negotiate, then the Eth-trunk1 ports split to many independent ports. The cisco switch's ports are independent ports also, then the loop appears, so it leads to the broadcast storm, which caused service break.
 Eth-Trunk1 is configured as lacp-dynamic mode
 
 The Eth-Trunk1 LACP negotiation failed, and the log records are as follows:
 
 After the LACP negotiation failed, the switch board Eth-Trunk1 members work in the independent mode. There is a MAC address flap between the ports. The log records are as follows:
 
 Eth-Trunk1 members have a network storm after configuring lacp-dynamic

Root Cause

Cisco switch doesn’t configure port group while E9000 switchboard that connected to them are Eth-Trunk. It leads to the mac flap & ping lost.

When configure LACP dynamic on E9000 Eth-trunk1, while cisco doesn’t configure port group, LACP dynamic failed to negotiate, then the Eth-trunk1 ports split to many independent ports. The cisco switch's ports are independent ports also, then the loop appears, so it leads to the broadcast storm, which caused service break.

Solution

Configure the Cisco switch ports that connecting to E9000 switchboard as port channel

Adapt the E9000 Eth-Trunk1 mode according to the configuration mode of the Cisco switch port channel

 

Laboratory verified docking mode

 

Work Mode

Cisco Port Channel Mode

Eth-Trunk Work Mode of E9000 switchboard

LACP

active

LACP static mode

Configured with command mode lacp-static

 

 

 

 

 

 

END