Sie sind auf Seite 1von 31

Preface

In this issue of ZTE's Maintenance Experience, we continue


to pass on various field reports and resolutions that are gathered
by ZTE engineers and technicians around the world.
The contents presented in this issue are twelve maintenance
cases of ZTE's data products.
Have you examined your service polices and procedures
lately? Are you confident that your people are using all the
tools at their disposal? Are they trained to analyze each
issue in a logical manner that provides for less downtime and
maximum customer service? A close look at the cases reveals
how to isolate suspected faulty or mis-configured equipment,
and how to solve a problem step by step, etc. As success in
commissioning and service is usually a mix of both discovery
and analysis, we consider using this type of approach as an
example of successful troubleshooting investigations.
While corporate leaders maintain and grow plans for
expansion, ZTE employees in all regions carry out with individual
efforts towards internationalization of the company. Momentum
continues to be built, in all levels, from office interns to veteran
engineers, who work together to bring global focus into their
daily work.
If you would like to subscribe to this magazine (electronic
version) or review additional articles and relevant technical
materials concerning ZTE products, please visit the technical
support website of ZTE CORPORATION (http://ensupport.zte.
com.cn).
If you have any ideas and suggestions or want to offer your
contributions, you can contact us at any time via the following
email: doc@zte.com.cn.
Thank you for making ZTE a part of your telecom experience!

Maintenance Experience
Bimonthly for Data Products
No.17 Issue 219, June 2010

Maintenance Experience
Editorial Committee
Director: Qiu Weizhao
Deputy Director: Chen Jianzhou
Editors:
Fang Xi, Wang Zhaozheng, Xu Xinyong,
Xiao Shuqing, Zhang Jian, Zhang Jiebin,
Zhou Guifeng, Zhao Cen, Zhao Haitao, Jiang
Haijun, Xu Zhijun, Huang Ying, Ge Jun, Dong
Yemin, Dong Wenbin
Technical Senior Editors:
Hu Jia, Tao Minjuan, Zhang Jianping
Executive Editor:
Zhang Fan

Maintenance Experience
Newsroom
Address: ZTE Plaza, No.55, Hi-tech Road

Maintenance Experience Editorial Committee


ZTE CORPORATION
June, 2010

South, Shenzhen, P.R.China


Postal code: 518057
Contact: Ning Jiating
Email: doc@zte.com.cn
Tel: +86-755-26771195
Fax: +86-755-26772236
Document support mail box: doc@zte.com.cn
Technical support website: http://ensupport.zte.
com.cn

Contents

LACP Configuration on ZXR10 8902 Switch.............................................................................................. 2


Sporadic Internet Access Failure................................................................................................................ 4
Experience About POS Interface Interconnection...................................................................................... 6
VRRP Changeover Failure......................................................................................................................... 9
Attentions for Configuring Out-of-Band and In-Band NM on UAS................................................................11
High CUP Usage Caused by Route Loop..................................................................................................14
Service Interruption after Capability Extension..........................................................................................17
Attention for MPLS Label Configuration.....................................................................................................19
Website Access Failure in MPLS VPN.......................................................................................................20
Access Failure Caused by MRU................................................................................................................23
VPLS Service Failure on T160G................................................................................................................25
Experiences About Flow Mirroring Configuration on ZXUAS 10800..........................................................27

June 2010

Issue 219

LACP Configuration on ZXR10 8902 Switch


Ma Xuetao / ZTE Corporation

Key words: ZXR10 8902, LACP, SmartGroup, active

LACP Principle
IEEE 802.3ad-based Link Aggregation
Control Protocol (LACP) is a protocol for
dynamic aggregation of links. LACP uses
Link Aggregation Control Protocol Data
Units (LACPDUs) to interact with peer
devices.
When LACP is enabled on an
interface, the interface will send

interface in a dynamic aggregation group.

Configuration Description
A ZXR10 8902 switch interconnects with its
peer device through the optical interface gei_1/23.
Th i s o p ti c a l i n te r fa c e i s i n V L A N 2 0 0 a n d
SmartGroup 1.
The LACP mode is in Active mode (it is

system MAC address, interface priority,

necessary to negotiate this parameter with the

port number and operation key. When

peer).

device compares the information with


the information that is saved on other
interfaces to select the interfaces that

Maintenance Experience

make an agreement on the joining or leaving of an

LACPDUs to advertise its system priority,

r e c e i v i n g t h e i n f o r m a t ion, the peer

can be aggregated. In this way, both devices can

Configuration Procedure
1. C r e a t e S m a r t G r o u p 1 a n d e n t e r
SmartGroup 1 configuration mode.

www.zte.com.cn

ZXR10-8902(config)#interface smartgroup1
ZXR10-8902(config-smartgroup1)#

2. Set the SmartGroup working mode to 802.3ad (LACP).


ZXR10-8902(config-smartgroup1)#smartgroup mode 802.3ad

3. Create the VLAN 200.


ZXR10-8902(config)#vlan200

4. Enter interface gei_1/23 and set the mode of this interface to Trunk.
ZXR10-8902(config)#interface gei_1/23

ZXR10-8902(config-gei_1/23)#switchport mode trunk

ZXR10-8902(config-gei_1/23)#switchport trunk vlan 200

5. Add the interface gei_1/23 to SmartGroup 1, and set the working mode to Active (it is
necessary to negotiate this parameter with the peer).
ZXR10-8902(config-gei_1/23)#smartgroup 1 mode active

After the configuration, users can view the interface state and LACP
state.

The configuration information on gei_1/23 is shown below.


ZXR10-8902(config)#show run interface gei_1/23
Building configuration
!

interface gei_1/23
out_index 25

hybrid-attribute fiber
negotiation auto

switchport mode trunk

switchport trunk vlan 200


switchport qinq normal
!

smartgroup 1 mode active

end

The LACP state is shown below.


ZXR10-8902(config)#show lacp 1 internal
Smartgroup:1

Flag *--LOOP is TRUE


Inter Interval
Actor
Port

Agg

State

Pri priority

Port

Inter

Port
Pri

Oper
Key

Port

State

RX

Machine

Mux

Machine

------------------------------------------------------------------gei_1/23

active

30

32768

0x111

0x3d

current

distributing

Data Products

June 2010

Issue 219

Sporadic Internet Access Failure


Qu Zhihui / ZTE Corporation

Key words: UAS 10600, address pool, route, VBUI

Malfunction Situation
The network topology of CNC in a city

The network phenomena are described as


follows:

is shown in Figure 1. The ZXR10 8512

1. There were always some users failing to get

switch was used for Layer 2 convergence.

online after dial-up and authentication. But they

PPPoE users were terminated on

could get online if they dialed up again or waited

UAS10600, and special line users were

for two minutes.

terminated on NE80E. Static back routes

2. When users fed back the problem, the

were configured on S8016 for the address

operation and maintenance personnel made some

pool configured on UAS 10600.

mock tests at the proximal end. There was no


problem.
3. The failure did not happen in any regular
pattern about time and geopolitical location.

Malfunction Analysis
1. Engineers checked the system configuration
and running state of UAS 10600. They found
no exceptions except that there were a lot of
broadcast packets received on the Layer 2 user
port connecting to 8512 per second.
2. To make sure that the large amount of
broadcast packets were not caused by network
factors such as virus, and to improve network
security, engineers configured an ACL on the 8512
Figure 1. Network Topology

Maintenance Experience

switch to prevent common viruses. After that, the

www.zte.com.cn

amount of broadcast packets received on UAS

was not UAS 10600. According to the

10600 did not decrease.

longest mask matching rule, the packets

3. To know the contents of the broadcast

to 192.168.211.0/24 were forwarded to

packets, engineers connected a PC to the 8512

another device instead of UAS 10600.

switch and transmitted all VLANs to this port

Therefore, the user who obtained the IP

transparently. So engineers could get the broadcast

address 192.168.211.1/24 failed to get

packet contents through the PC. According to the

online.

information in the packets, besides the proper

8. If the user disconnected the

broadcast packets, there was a user scanning (the

connection and dialed up again, according

traffic was very little). Therefore, the failure was not

to the principle of poll-and-allocate address

caused by broadcast packets.

in the address pools on UAS 10600, it

4. The next day, engineers found that there

was possible for the user to obtain an IP

was a user sending more than 900 ARP broadcast

address in another segment. So the user

packets per second. Engineers checked the

could get online.

running of UAS 10600, and they found that the

The user could get online after waiting

device was running properly. Therefore, this was

for two minutes, because the user dialed

not the cause of the failure.

up again to obtain an IP address in

5. To reproduce the failure, engineers dialed up

another segment during this period.

and got offline repeatedly to conduct tests. At last,

The operation and maintenance

they found that the PC failed to get online when the

personnel made some mock tests at

PC passed the dial-up authentication and obtained

the proximal end but they did not find

the IP address 192.168.211.185. The PC could

any problem. This was because the IP

ping through the interface address on UAS 10600,

address obtained by the PC was not in the

but it could not ping through the interface address

segment 192.168.211.1/24.

on the peer device 8512. Therefore, engineers


considered that there was a fault about the route in
this segment.

Solution
Engineers informed the operation

6. There were 10 VBUI interfaces (there

and maintenance personnel to check the

were 10 address pools in total) on UAS 10600.

routing tables. It was found that the route

There were two Class-C addresses on each

to the segment 192.168.211.1/24 was

VBUI. The IP address of the VBUI to which the

advertised by the S8016. When the route

IP address obtained by the PC belonged was

was deleted, the fault did not occur any

192.168.210.1/23.

more.

Engineers checked the routes on the 8512


switch, as shown below.

According to the analysis, users


obtained the IP address with the incorrect

7. On the 8512 switch, the next hop address


of the route to the address 192.168.211.0/24

route, which caused the Internet access


failure.

Detination/Mask

Protocol

Pre

Cost

Nexthop

Interface

192.168.210.0/23

OSPF

10

11

10.10.239.146

Vlan-interface 4071

192.168.211.0/24

O_ASE

150

1001

10.10.239.137

Vlan-interface4096

Data Products

June 2010

Issue 219

Experience About POS Interface


Interconnection
Liu Peng / ZTE Corporation

Key words: POS, GER, CISCO

Malfunction Situation
On a network, a ZTE GER router

command was used to check the port state, the


following result was displayed.

interconnected with a CISCO 6509 through

GER04#show interface pos3_2/1

POS interfaces. The network topology is

pos3_2/1 is up,

shown in Figure 1.

line protocol is down

The result shows that the link layer protocol is


down, and the connection cannot be established.

Malfunction Analysis
Figure 1. Network Topology

On the POS interfaces of GER04,


only IP addresses were configured. Other
parameters used the default values. The
configuration is shown below.
interface pos3_2/1
ip address

192.168.10.30

255.255.255.252

When the show interface pos3_2/1

Maintenance Experience

When POS interfaces are used for


interconnection, there are some parameters
that must be considered. The following contents
describe these parameters in details according to
the configuration on the GER router.

clock source {internal | external | line}


This command is used to configure the line

clock. The keyword internal means to extract the


clock of the device its own. The keyword external
means to extract the clock from SMP. The keyword
line means to extract the clock from the line.

www.zte.com.cn

If two routers interconnect with each other

identify the type of load encapsulated in

back-to-back through POS interfaces, the clock of

SONET FOH. The range of the value is

one router should be set to internal, and the clock

0 ~ 255. Different values have different

of the other router should be set to line. By default,

meanings, as listed in Table 1(decimal is

the clock on POS interface of ZTE GER is internal.

used during configuration, and the value

crc {16 | 32}

listed below is in hexadecimal).

This command is used to configure the CRC

According to the list, if the scrambling

mode on a POS interface. The CRC mode can

function is enabled, the C2 value on the

be set to 16 bits or 32 bits. When two routers

POS interface should be set to 0X16. If

interconnect with each other through POS

the scrambling function is disabled, the C2

interfaces, the CRC modes of the interfaces should

value on the POS interface should be set

be consistent. By default, the CRC mode on ZTE

to 0XCF. By default, the C2 value on the

GER is 32.

POS interface of ZTE GER is 0X16, which

scramble { payload-disable | payload-enable}

corresponds to the scrambling function

This command is used to enable or disable

enabled on the POS interface.


Note: As the system of ZTE GER does

the scrambling function. By default, this function is


enabled on ZTE GER.

not support non-scrambling function in an

flag C2 <number>

SDH header, the C2 value 0XCF does not

C2 is the path signaling label. It is used to

take effect.

Table 1. SONET Payload Contents

Hex Value

SONET Payload Contents

flag J0 <number>
J0 belongs to the SOH byte. It is used

to detect the continuity of the connection

00

Unequipped.

01

Equipped-non-specific payload.

02

Vitual Tributaries (VTs) inside (default).

of different vendors. For example, devices

VTs in locked mode (no longer

of CISCO use 1-byte integer mode,

supported).

devices of HW use 16-byte character

04

Asynchronous DS3 mapping.

string mode, and devices of ZTE also use

12

Asynchronous DS-4NA mapping.

1-byte integer mode.

03

13
14
15
16

in section between two interfaces. The


configurations may be different on devices

Generally, the value of this field should

Asynchronous Transfer Mode (ATM)


cell mapping.

be consistent on both ends, especially

Distributed Queue Dual Bus (DQDB)

when this field is matched strictly on

cell mapping.

CISCO devices. Usually, this parameter is

Asynchronous Fiber Distributed Data

set to 0X01 on ZTE devices, which means

Interface (FDDI) mapping.

compatible. If there is any problem about

IP inside Point-to-Point Protocol (PPP)


with scrambling.

interconnection, it is also possible to set


the value to 0X01 on the peer. By default,

CF

IP inside PPP without scrambing.

the value of this field is 0xCC on the POS

E1-FC

Payload Defect Indicator (PDI).

interface of ZTE GER.

Test signal mapping (see ITU Rec.

FE
FF

G.707).
Alarm Indication Signal (AIS).

flag S1S0 <number>


S1S0 is the line synchronization state

label to set the synchronization state on

Data Products

June 2010

Issue 219

the line terminals. The range is 0 ~ 3. This

of the POS interfaces on the two devices. They

value is not defined in SONET standard

found that there was a parameter that must be

and it is not checked at the receiving side.

paid attention to, that is, the encapsulation type on

However, this value is defined in SDH

the POS interface. By default, the encapsulation

standard and it is checked at the receiving

mode on the POS interface of ZTE GER is PPP,

side. If the value is not equal to 2 at the

which cannot be modified. However, the default

receiving side, the interconnection will

encapsulation mode on the POS interface of

fail. By default, the value is 2 on the POS

CISCO device is HDLC.

interface of ZTE GER, which means SDH


frame format.

Engineers modified the encapsulation mode


on the POS interface of the CISCO device to PPP.

framing {sonet | sdh}

The link layer protocol was up, and the two devices

This command is used to configure

could ping through each other successfully.

the frame format on POS interfaces. The


frame format should be consistent on both

Summary

ends. If routers are connected through a


transport device, the frame format should

The default parameter values on the devices of


ZTE, HW and CISCO are compared in Table 2.

be consistent with that on the transport

During the configuration of POS interfaces,

device at the same time. By default,

do pay attention to the related parameters at both

the frame format is sonnet on the POS

sides, especially pay attention to the encapsulation.

interface of ZTE GER.

The configuration of encapsulation should be


consistent at both ends.

Solution

As the default encapsulation mode on the POS

In this malfunction case, even when all

interface of ZTE GER is PPP and it cannot be

parameters were set to keep consistent

configured, make sure that the encapsulation mode

with those of the peer CISCO router,

is set to PPP on the POS interface of the peer

the link layer protocol state on the POS

device when the two devices are interconnected

interface was still not up.

through POS interfaces. Otherwise, the link layer

Engineers checked the configuration

protocol cannot be up.

Table 2. The Devices of ZTE, HW and CISCO

ZTE

Maintenance Experience

Huawei

CISCO

mtu

1500

1500

4096

CRC

32

32

16

Scrambling

scrambling

scrambling

No scrambling

C2

0x16

0x16

0xCF

S1S0

J0

0xCC

Clock source

Internal

Framing

Sonet

Encapsulation

PPP, cannot be configured HDLC

Adjusted automatically, cannot


be configured
NetEngine
Master (corresponding to
internal)
SDH

2
0xCC
Line
SDH
HDLC

www.zte.com.cn

VRRP Changeover Failure


Wu Lifeng / ZTE Corporation

Key words: T64E, VRRP, Track, firewall

Malfunction Description
VRRP ran between the routers and between
the firewalls. The devices interacted with VRRP
protocol packets through the specific heart
beaten line interfaces and were not affected by
the different segments on the heart beaten line
interfaces. When the optical fibers between the
Juniper firewalls and the T64E routers were
single-pass, as shown in Figure 1, the service
communication between the client connecting to
the firewalls and the server was interrupted.

Interface

Own Pre State


Group-addr
gei_1/4.12
N

Grp Pri Time


Master-addr

12

10.134.64.155
gei_1/3.3
N

gei_1/4.13
Y

13

gei_1/4.11
Y

gei_1/3.2
Y

gei_1/3.1
Y

130 14.492

130 14.492

130 14.492

Master 172.16.1.161

172.16.1.163
N

11

Master 10.7.29.153

10.7.29.155
N

130 8.492

Master 192.168.1.153

192.168.1.155
N

130 14.492

Master 192.168.1.161

192.168.1.163
N

130 14.492

Master 10.134.64.153

Master 10.7.29.161

10.7.29.163

2. The result showed that the VRRP


state was normal. Engineers pinged
the interface addresses on the Juniper
firewalls, but the addresses could not be
Figure 1. Network Topology

Malfunction Analysis
1. Engineers logged into R1 to check the VRRP
state, as shown below:
R1#show vrrp brief

Interface total: 12, Group total: 12,

Master total: 9

pinged through, as shown below:


R1# ping vrf juniper-mss
10.7.29.164

sending 5, 100-byte ICMP echoes


to 10.7.29.164, timeout is 2
seconds.
.....

Success rate is 0 percent(0/5).

Data Products

June 2010

Issue 219

3. Engineers checked FW-1. They

are interconnected with devices of other vendors,

found that the port on FW-1 was down.

usually the working mode of the ports is set to

At present the VRRP state was Backup.

mandatory mode. Otherwise, the interconnections

Engineers checked the problem and found

will fail. However, in this way, single pass through

that FW-1 did not receive optical signal

optical fiber is unavoidable. It is possible to solve

from R1. The reason was that, as the

the problem in the following methods:

working mode of the interconnected ports

1. One method is to configure a VRRP group to

on R1 and FW-1 was mandatory 1000 M,

trace the link state of a track. If the interface state

as long as R1 received optical signal, the

becomes down from up, the priority of the interface

port would be up and VRRP changeover

decrements. If the interface state becomes up from

would not be implemented.

down, the priority of the interface increments to

R1#show interface gei_1/3


gei_1/3 is up,
up

line protocol is

Description is TO_XHP-JUNIPER1_gei_1/1

Solution

speed up the election of VRRP Master router.


In fact, the working principle of a track is also
based on the up/down working mechanism of a
port. Therefore, this method is not feasible.
2. The second method is the IP address
based detection mechanism. At present, Juniper
firewall supports this protocol. Even when a fault

Engineers disconnected the other

occurs on the port, the state displayed is still UP.

optical fibers on the Juniper firewalls. The

As long as the address of this port cannot be

services were recovered.

pinged successfully, VRRP changeover will not be

Summary
For VRRP routers, only the Master
router is responsible for forwarding the

implemented.
3. Another method is to add a switch. Figure
2 shows a typical network architecture. It is
recommended to use this method.

packets with address that is associated


to the virtual router and responding to the
ARP requests for this IP address. In the
situations such as priority is not configured
manually, VRRP is implemented according
to the state of ports.
In this malfunction case, the VRRP
state on the routers was not changed over.
However, FW-1 found that the uplink was
interrupted, so FW-1 changed over all
services to FW-2 to send the services to
R2. As R2 was the Backup VRRP router,
it would not handle with the ARP requests.
Therefore, the services were interrupted.
In existing network, when ZTE devices

10

Maintenance Experience

Figure 2. Typical Network Architecture

In brief, when there was the switch (shown in


Figure ) on the network, conduct tests about VRRP
hardware handover and VRRP logical changeover.

www.zte.com.cn

Attentions for Configuring Out-of-Band


and In-Band NM on UAS
Liu Peng / ZTE Corporation
Key words: UAS, device management, in-band management, out-of-band management,
Netnumen

Network Topology
The broadband access network of a service
provider used ZTE GER/UAS10600/T40G/
FASP9800 network structure. The network topology

is shown in Figure 1. It was required


to manage the devices through private
network addresses.
After communicating with the office

Figure 1. Network Topology

Data Products

11

June 2010

Issue 219

personnel, engineers combined in-band


network management and out-of-band
network management to build the network
for network management.

For UAS devices and GER routers,


engineers configured private network
addresses through out-of-band
management interface to connect the
devices to VPN and then interconnect
them to the Netnumen Server.

For T40G switches and the downlink


devices, as there were quire many
devices, it was impossible to provide
an individual out-of-band management
link for each device. Therefore,
engineers used in-band network

management:
interface vbui11

ip address 192.168.10.1 255.255.255.0


ip host 192.168.10.2 slot 1 port 3
vlan 199

/*Add the network addresses of T40G

and downlink devices in IP host mode*/


ip host 192.168.10.3 slot 1 port 3
vlan 199

Configuration related to out-of-band network


management:
nvram mng-ip-address 192.168.10.4
255.255.255.0

There was no information indicating

port on each T40G switch, thus

alarms or errors during the configuration. If the

connecting the devices to VPN after

configurations were feasible, it was only necessary

aggregation and then interconnecting

to configure network management addresses in

them to the Netnumen Server.

192.168.10.0/24 segment for T40G switches and

H o w e v e r, i t w a s o n l y p o s s i b l e

downlink devices in the network.

to manage the devices through the

After the configuration, engineers checked

Netnumen Server or the VPN network in

whether the devices were reachable on UAS, as

out-of-band network management mode.

shown below:

Server, or when the VPN was unavailable,


engineers could not manage T40G
switches and the downlink devices.
For security, engineers configured
Layer 3 interfaces on T40G switches
and UAS devices. Therefore, once the
out-of-band network management was
unavailable, engineers could telnet to
T40G switches and the downlink devices
through UAS devices. In this way, it is
possible to manage all devices through the
private network, which was convenient for
routine maintenance and remote diagnosis
when faults occurred on the devices.
To make the configuration as simple
as possible, engineers configured the

Maintenance Experience

Configuration related to in-band network

management mode through a service

Once a fault occurred on the Netnumen

12

following commands on UAS.

ZXUAS#Ping mng 192.168.10.2

/*ping

the management address of T40G on the


out-of-band management interface*/
sending 5,100-byte ICMP echos to

192.168.10.2,timeout is 2 seconds.
!!!!!

Success rate is 100 percent(5/5),roundtrip min/avg/max= 0/0/0 ms.


ZXUAS#ping 192.168.10.2

/*

ping the management address of T40G on


the in-band management interface*/
sending 5,100-byte ICMP echos to

192.168.11.2,timeout is 2 seconds.
!!!!!

Success rate is 100 percent(5/5),roundtrip min/avg/max= 0/4/20 ms.

www.zte.com.cn

The information showed that T40G could be

segment as the address segment of T40G

pinged successfully through in-band and out-of-

and the downlink devices for in-band

band management interfaces. As the out-of-band

management through UAS. The address

management interface of UAS and the specific

segment was 192.168.11.0/24. That is,

management interface of T40G were connected

T40G and the downlink devices used

to the same HUB and they were in the same

segment 192.168.10.0/24 as the out-of-

segment, T40G could be pinged successfully

band network management addresses,

through the out-of-band management interface.

and used segment 192.168.11.0/24 as the

Meanwhile, engineers added the addresses of

in-band network management addresses.

T40G and the downlink devices by setting IP

After the modification, engineers

host through VBUI11 interface on UAS, so T40G

checked the states on the devices. There

could be pinged successfully through the in-band

was no alarm.

management interface.

Malfunction Situation

Summary
On UAS 10600 or UAS 10800E,

If users used the in-band address 192.168.10.1

no stipulate defines that the in-band

as the network management address of UAS

management address and the out-of-band

and then added an NE through Netnumen,

management address cannot be in the

alarms indicating that the NE was disconnected

same segment. In practical configurations,

periodically occurred on the NE.

the two addresses can even be the same

Meanwhile, if users used the in-band address

IP address.

192.168.10.4 as the network management address

To use in-band management and out-

of UAS and then added an NE through Netnumen,

of-band management at the same time,

the NE state was steady and the connection was

it is necessary to avoid using the in-band

normal.

management address and the out-of-band

Solution

management address that are in the same


segment and to ping through each other

In this special network management structure,

successfully. Users can solve such problems

engineers used the following method to solve the fault.

through physical isolation or by dividing the

Engineers used another private network

two addresses in different segments.

Data Products

13

June 2010

Issue 219

High CUP Usage Caused by Route Loop


Fu Tao / ZTE Corporation

Key words: ICMP, route loop, CPU usage

Malfunction Situation

there were packets lost. When they pinged remote

Internet bar users said that when they


pinged the gateway switch ZXR10 3952,

websites, the delay was normal and download was


normal.

the delay was long, with the average delay

The network topology is shown in Figure 1.

about more than 200 ms, and sometimes

Figure 1. Network Topology

Malfunction Analysis

address of the Alcatel router. The delay was about

1. Engineers logged in to ZXR10 3952

200 ms.

and pinged the user address to conduct

Engineers connected a PC to the switch and

tests. They found that the delay was long,

pinged the Alcatel router, the delay was also about

with the average delay about 202 ms. The

200 ms.
Engineers executed show process command to

maximum delay was 300 ms.


Engineers pinged the interconnection

check the CPU usage. It was 22%, as shown below:

ZXR103952#show process
M: Master processor
S: Slave processor

Peak CPU: CPU peak utility measured in 2 minutes


PhyMem: Physical memory (megabyte)
Memory
MP(M)

34.662%

Panel
1

CPU(5s) CPU(30s) CPU(2m)


22%

22%

2. Engineers enquired about the CPU


usages on other ZXR10 3952 switches.

14

Maintenance Experience

22%

Peak CPU
33%

PhyMem
256

Buffer
0%

the same traffic. Obviously, the CPU usage of this


XR10 3952 switch was much higher.

They were told that the CPU usages were

Engineers executed the show logging alarm

about 10% on other switches with about

command but they did not find any alarms for

www.zte.com.cn

exceptions such as ARP attack and so on. They

that the count of badhop increased quickly

executed the show ip traffic command and they found

during a short period, as shown below:

ZXR103952#show ip traffic
IP statistics:

Rcvd: 1376900 total, 200638 local destination


format errors

checksum errors

unknown protocol
0

Frags:reassembled
0

timeouts

bad hop count


29614

couldn't reassemble

3. Engineers checked the switch and they found

delay, and even some packets were lost.

that there was only one default route directing to

User could download and ping websites

the Alcatel router. Therefore, engineers doubted

properly, because those packets might be

that the high CPU usage was caused by too many

only forwarded instead of being handled

packets with TTL=1 sent to CPU due to a route

with by the CPU on the ZXR10 3952

loop. As a result, the CPU was busy handling with

switch.

these packets with TTL=1.

To validate the surmise, engineers

As the CPU usage was high, the ping packets


of the Internet bar users were handled with after a

executed the debug ip icmp command on


the switch. The result is shown below:

05:17:27

IP

ICMP:sent

type

time

exceeded,

code

0,

src

05:17:28

IP

ICMP:sent

type

time

exceeded,

code

0,

src

05:17:29

IP

ICMP:sent

type

time

exceeded,

code

0,

src

05:17:30

IP

ICMP:sent

type

time

exceeded,

code

0,

src

05:17:40

IP

ICMP:sent

type

time

exceeded,

code

0,

src

172.16.237.118,dst 123.55.214.19
172.16.237.118,dst 192.168.33.54
172.16.237.118,dst 192.168.33.54
172.16.237.118,dst 123.55.214.19
172.16.237.118,dst 58.215.81.76

4. The result showed that the ZXR10 3952

debug ip source 192.168.33.54 command

switch received ICMP packets with TTL=1 from

(take the IP address 192.168.33.54 as an

123.55.214.19 and 192.168.33.54. As the TTL

example) to view the detailed information,

values of these packets were 1, these packets were

as shown below:

sent to the CPU for handling. Engineers executed


ZXR103952# debug ip source 192.168.33.54
IP source debugging is on
WeiXiaoXiJie3952#

05:40:14 IP source:192.168.33.54 forward src 192.168.33.54(vlan2),dst

Data Products

15

June 2010

Issue 219

10.10.74.9(vlan2)

05:40:14 IP source:192.168.33.54 forward src 192.168.33.54(vlan2),dst


10.10.74.9(vlan2)

05:40:14 IP source:192.168.33.54 forward src 192.168.33.54(vlan2),dst


10.10.74.9(vlan2)

05:40:14 IP source:192.168.33.54 forward src 192.168.33.54(vlan2),dst


10.10.74.9(vlan2)

5. The information showed that the

10.10.74.9 belonged to by executing show ip

destination of the packet with the source

route connected command on the ZXR10 3952

address 192.168.33.54 was 10.10.74.9.

switch. Therefore, the packet was sent back to the

This packet could reach the ZXR10

router according to the default route.

3952 switch, so there must be a route

In this way, a loop was formed between the

heading for the segment which 10.10.74.9

ZXR10 3952 switch and the router. To validate the

belonged to directed to the ZXR10 3952

surmise, engineers traced the destination address

switch. However, engineers did not find

10.10.74.9 on the ZXR10 3952 switch, as shown

any direct-connected segment which

below:

ZXR103952# trace 10.10.74.9

tracing the route to 10.10.74.9

1 172.16.237.117

260 ms

220 ms

2 172.16.237.118

240 ms

240 ms

on the router*/

interconnecting to the router*/


3

According to the analysis, engineers


found that there was a route loop. There
were three methods to solve the problem.
1. Configure an extended ACL to filter

/*address of the interface

on the router, engineers checked the CPU usage


again. The CPU usage and the delay of ping
packets were normal.

Summary

the addresses that were unreachable on

When the device receives a packet with

the uplink interface of the ZXR10 3952

TTL=1, it will send an ICMP time-out packet to the

switch

remote source address. The source address of this

2. Configure black hole routes directing

time-out packet is an address on this device, and

to the unreachable segments on the

the destination address of this packet is the remote

ZXR10 3952 switch

source address.

3. Delete the useless static route


configuration on the router
The third method met the requirements
of network configuration specifications.
After deleting the useless static routes

Maintenance Experience

240 ms

/*address of the interface

Solutions

16

220 ms

The debug ip source <ip-address> command can


be used to view the related information in detail of
the packet with this IP as its source address, such
as the destination address, and so on.

www.zte.com.cn

Service Interruption after Capability


Extension
Liu Jiangdong / ZTE Corporation

Key words: HSRP, MAC drafting, group number

Malfunction Situation
The network topology is shown in Figure 1. As

were assigned to the RNSc and BSCs in


machine room 2. After the changeover,

some new 2G sites and 3G sites of an office were


added, the loads of RNCs and BSCs in the original
machine room were not enough. It was necessary
to extend the capabilities of a device group in
machine room 2. After the capability extension, the
coupling from some 2G and 3G base stations to
the RNCs and BSCs was intermittent.

Malfunction Analysis
The two CISCO 6513 in machine room 1
enabled HSRP protocol to work as the gateway
of the base stations. T64G only executed Layer 2
transparent transmission. Services run properly
before the capability extension.
After the capability extension, the group of
the devices in machine room 2 also used HSRP
protocol to work as the gateway. Some sites

Figure 1. Network Topology

Data Products

17

June 2010

Issue 219

however, the communications of the sites


assigned to the two machine rooms were
After the malfunction occurred,
engineers took the following steps:
1. Engineers checked whether there
was any alarm information on CISCO6513.
But they did not find any exception.
2. Engineers checked the routes to
RNCs and BSCs on CISCO6513. They
found that static route load sharing
was configured from RNCs and BSCs
to CISCO6513. The static route load
sharing entry was proper. Engineers could
ping RNCs and BSCs successfully on

standby 1025 preempt delay minimum

10

standby 1025 authentication md5

key-string HSRP

The related configuration on CISCO6513 in


machine room 2 is shown below:
interface Vlan25
description

###connected

T64G_2_GEI_1/18###
ip

address

255.255.255.0

to

10.10.25.252

standby version 2

standby 1025 ip 10.10.25.254

CISCO6513.
At the base station side, CISCO6513
connected to the base stations through
direct-connected segments. They could
ping each other successfully. Engineers
checked the ARP entries. There were ARP
entries to the base stations, but they were
not steady and were learnt repeatedly and
discontinuously. Therefore, there were
some problems.
3. Engineers checked ZXR10 T64G
switches. They found that there were a lot
of alarms indicating MAC address drifting.
The MAC address was the virtual MAC of
the HSRP gateway on the base stations.
4. Engineers checked the HSRP
configuration of Layer 3 VLAN on
CISCO6513.
The related configuration on
CISCO6513 in machine room 1 is shown
below:

standby 1025 timers 2 5

standby 1025 priority 110

standby 1025 preempt delay minimum

10

standby 1025 authentication md5

key-string HSRP

According to the configuration, the standby

group numbers were the same. According to


the information provided by CISCO, the last
several bits of the virtual MAC address in HSRP
are calculated according to the group number.
Therefore, the same HSRP group resulted in the
same MAC address. As a result, MAC drafting
occurred on ZXR10 T64G switches in this network
environment, and the fault occurred.

Solution
Engineers modified the related configuration
of the HSRP group number on the same Layer
3 interfaces. The services were recovered. The
coupling from some 2G and 3G base stations to

interface Vlan25

the RNCs and BSCs was proper. The ARP entries

description ###connected to

on CISCO6513 were steady.

ip

VRRP, it is also necessary to avoid using the same

T64G_1_GEI_1/18 ###
address

255.255.255.0

10.9.25.252

standby version 2

Maintenance Experience

standby 1025 timers 2 5

standby 1025 priority 110

intermittent.

18

standby 1025 ip 10.9.25.254

We can associate VRRP from CISCO HSRP. In


VRRP group numbers in such network structures.

www.zte.com.cn

Attention for MPLS Label Configuration


Wang Feng / ZTE Corporation

Key words: ACL, MPLS, label, loopback address, FEC

Malfunction Situation
At present, for the VPN configuration on ZXR10
T128/T64E of Version 2.6.02 or previous versions,
usually ACL is configured to control the generation
of labels, as shown below:
mpls ip

mpls ldp router-id loopback1


mpls ldp access-fec for 1

distribute labels.
The new label is ready, but restarting
the MPLS process will interrupt the
services.
Therefore, it is necessary to configure
mpls ldp access-fec host-routeonly command during the MPLS VPN
configuration.
This command is to generate labels

acl standard number 1

for host-routes only, and not to generate

0.0.0.0

or interconnected segments between

0.0.0.0

the labels distributed for the loopback

rule 1 permit 192.168.66.30

labels for routes of service segments

rule 2 permit 192.168.98.27

devices. In this way, for VPN services,

rule 3 permit 192.168.128.119

addresses of the PE devices are enough.

rule 4 permit 192.168.128.118

distributed for the loopback addresses

0.0.0.0
0.0.0.0

rule 5 permit 192.168.128.117

0.0.0.0

The advantage is that the number of the labels


generated by the device can be controlled. This

However, unnecessary labels may be


of the devices that do not work as PEs.
When there are a lot of devices on the
entire network, many labels will be wasted.

Solution

saves the valuable memory resources of the device

During the MPLS configuration on

effectively. However, there is a trouble under the

ZXR10 T128/T64E of Version 2.6.02 or

situation when it is necessary to distribute a label

previous versions,

for the loopback address of a new added PE

on the network, it is recommended

device.

to use the configuration that only

In such a situation, is it only necessary to re-

distributes labels for host-routes.

compile the ACL? If so, ACL is changed, but the


number of labels does not increase, and the label

When there are not too many devices

When there are many devices on the

of the new added PE cannot be viewed. Therefore,

network, but the number of PE devices

it is also necessary to restart the MPLS process

is steady (that is, PE devices will not be

to make it use the ACL that has been modified to

added randomly), it is recommended

Data Products

19

June 2010

Issue 219

to use ACL to control the generation of

any more. It is only necessary to execute the

labels.

mpls ldp access-fec force command after ACL

It is recommended to use one of the

modification. The new ACL will take effect at once.

methods when MPLS is configured.

This procedure will not affect the running services.

A new function is added on ZXR10

At present, this function has been applied on

T128/T64E of Version 2.6.03B, that is,


FEC takes effect immediately.

current networks. It has good effect.


Therefore, it is recommended to use the ACL-

When the ACL-based policy is

based policy to control the number of labels

configured to restrict the number of labels,

distributed to devices when MPLS is configured

if the ACL needs modification, it is not

on ZXR10 T128/T64E of Version 2.6.03B or later

necessary to restart the MPLS process

versions.

Website Access Failure in MPLS VPN


Qu Zhihui / ZTE Corporation

Key words: MPLS, VPN, MTU, fragment, WWW, label

Malfunction Situation

departments. The VPNs communicated with the

On the government network of a

VPN of province level. On the previous network,

city, MPLS/VPN technology was used to

CISCO7206 worked as the PE device and

construct VPNs for different administration

established MP-BGP neighbor relationship with


the PE device (CISCO GSR12000) in the VPN of
province level.
To improve the network performance and
construct the MPLS VPN network covering levels
of province, city and county, ZXR10 T64G was
used as the PE device of city level. It established
MP-BGP neighbor relationship with the PE device
of province level. Meanwhile, ZXR10 T64G worked
as the route reflector and established MP-BGP
neighbor relationship with the PE devices of county
level.
The government network topology before

Figure 1. Government Network Topology Before Update

20

Maintenance Experience

update is shown in Figure 1.

www.zte.com.cn

The government network topology after the


update is shown in Figure 2.

5. According to the information,


the WWW server in Site A replied with

After the government MPLS/VPN network

some packets. So there was no problem

was rebuilt, users in Site B failed to access the

about the routes in the VPN. Engineers

homepage of the WWW server in Site A.

found that the fragment field in the IP

Malfunction Analysis
1. Engineers checked the devices. The fault did
exist.

packets were set to fragment not allowed.


Engineers inferred that the network card of
the WWW server was set not to fragment
the packets.

2. Engineers could not ping the WWW server in


Site A successfully from the PC in Site B. But they
could ping the VRF interface address of VPN1 in
Site A successfully.
3. Engineers connected the CE device in Site
B to CISCO7206 again. Then engineers could
not ping the WWW server in Site A successfully
from the PC in Site B. But they could access the
homepage of the WWW server. Therefore, ICMP
filter function was enabled on the network in Site A.
There might be no problem about the routes in the
VPN.
4. As it is impossible to find out the problem
with the ping tool, engineers captured packets to
obtain more information. They connected the CE
device in Site B to ZXR10 T64G again. Engineers
captured packets with SNIFFER software when the
PC in Site B accessed to the WWW server in Site A.
The packet information is shown in Figure 3.

Figure 2. Government Network Topology After Update

Figure 3. The Packet Information

Data Products

21

June 2010

Issue 219

The maximum length of a proper HTTP


packet is 1500 bytes. As the packet was

Maintenance Experience

as described below:

transmitted in MPLS VPN, after it was

1. Before the network update, in normal

encapsulated with two layers of MPLS

situations, the HTTP request packets or the TCP

labels (in the situation of penultimate hop

acknowledge packets sent to the WEB server in

popping, the external label was popped up,

Site A by the PC in Site B were small packets.

and only one VPN label left), the length of

Even if the packets were encapsulated with two

the packet was more than the default MTU

layers of MPLS labels, the length was still less than

(1500 bytes). If the devices on the MPLS

1500 bytes. Therefore, there was no problem if

network did not support to transmit jumbo

the interconnected interfaces on CISCO7206 and

frames and the packet was not allowed

CISCO GSR used the default MTU.

to be fragmented, the packet would be

2. Generally, the length of the HTTP packets

dropped. An ICMP packet would be sent

sent to the PC in Site B by the WEB sever in

to the source of this packet, indicating that

Site A was 1500 bytes. After the packets were

the device received a packet of which the

encapsulated with two layers of MPLS labels, the

length was more than the MTU but the

total length was more than the default MTU (1500

packet was not allowed to be fragmented.

bytes) consequentially. The CISCO GSR device

Solution

22

according to the traffic model of the HTTP access,

supported to forward jumbo MPLS frames. When


CISCO7206 received the jumbo MPLS frames, as

Preliminarily, it was considered that the

it worked as the PE device, the packets forwarded

fault was caused as the MTU of the path in

on the VRF interface were common IP packets

the MPLS VPN was smaller than the size

and the MPLS header information was removed.

of the MPLS frame. Which was the device

Therefore, the length was not more than 1500

with the MTU bottleneck in the MPLS

bytes. As a result, the PC in Site B could access

domain?

the WEB server in Site A.

The NP chip of ZXR10 T64G supported

3. After the network update, ZXR10 T64G

to forward jumbo MPLS frames, therefore,

worked as the PE device, and CISCO7206 worked

the fault was not caused by ZXR10 T64G.

as the P device. If the length of an HTTP packet

As the CISCO GSR device also supported

sent to Site B from Site A was 1500 bytes, when

to forward jumbo MPLS frames, the

this packet was forwarded to CISCO7206, as

fault might be caused by CISCO7206.

ZXR10 T64G enabled penultimate hop popping, the

Engineers informed the agent to modify

packet forwarded to ZXR10 T64G by CISCO7206

the MTU on the interface of CISCO7206 to

contained an MPLS label. That is, the VPN label.

1544. The fault was solved.

As a result, the length of this packet was more than

Why could Site B access the WEB

the default MTU (1500 bytes), and so the packet

server in Site A before the network

was dropped. This was why the PC in Site B could

update? Engineers analyzed the reason

not access to the WEB server in Site A.

www.zte.com.cn

Access Failure Caused by MRU


Wu Lifeng / ZTE Corporation
Key words: MPLS, VPN, MRU, MTU, GER

Network Topology
As shown in Figure 2, ZXR10 T1200 worked as
a P device. ZXR10 T64E worked as a PE device,
and users connecting to ZXR10 T64E could access
Internet properly. GER worked as a PE device, and
users connecting to GER could only open baidu. They
could not open other portals, such as sina, sohu, and
so on.

Engineers could ping baidu successfully

on PC2, but they could not ping sina and sohu


successfully.
2.

echos to 10.6.0.13,timeout is
1 seconds.

...........

Success

percent(0/11).

rate

is

3. According to the above steps,


engineers affirmed that there was no

Malfunction Analysis
1.

sending 20,1493-byte ICMP

Engineers pinged the interface address of

PING NE40 on GER, as shown below:

problem about the VPN routes. The problem


may be the MTU parameter configuration
on the downlink interface of ZXR10 T1200
connecting to GER.
Engineers checked the configuration on
the interfaces of ZXR10 T1200 connecting
to T64E and GER. They found that the MTU

GER-1#ping vrf vpn-mss 10.6.0.13

was 1600 and the MRU was 7600 on ZXR10

NE40*/

was 1700 on T64E. The MTU was 1500 and

/*the address of the PE device


sending 5,100-byte ICMP echos to
10.6.0.13,timeout is 2 seconds.
!!!!!

T1200. The MTU was 1600 and the MRU


the MRU was 1500 on GER.
MRU indicates Maximum Receive Unit.

Success rate is 100 percent(5/


5),round-trip min/avg/max= 0/8/20
ms.

GER-1#ping vrf vpn-mss 10.6.0.13


option 20 1492 1

sending 20,1492-byte ICMP echos to


10.6.0.13,timeout is 1 seconds.
!!!!!!!!!!!!!!!!!!!!

Success rate is 100 percent(20/

20),round-trip min/avg/max= 0/5/20


ms.

GER-1#ping vrf vpn-mss 10.6.0.13


option 20 1493 1

Figure 2. Network Topology

Data Products

23

June 2010

Issue 219

It is the maximum length of a packet that can

The forwarding intermediate nodes did not support

be received. If the length of a packet is more

fragment, so the jumbo packets were dropped directly.

than the MRU, this packet will be dropped.

3. At the GER side: On the incoming direction, it

MRU effects on the ingress interface when

was only necessary to check the MRU. Generally, the

the device receives traffic, that is, MRU is

length of an HTTP packet sent from the Internet to

uplink-valid.

PC2 was 1500 bytes or even larger. After the packet

MTU indicates Maximum Transit Unit.

was encapsulated with two layers of MPLS labels,

It is the maximum length of a packet that

the total length of the packet was consequentially

can be forwarded. If the length of a packet

more than the default MRU (1500 bytes). The MRU

is more than the MTU, the packet will be

on the interface of GER was 1500 bytes. Therefore,

dropped or fragmented. MTU effects on the

the packets with length more than 1500 bytes were

egress interface when the device forwarding

discarded.

traffic, that is, MTU is downlink-valid.


Engineers analyzed the data forwarding
flow, as described below:

total length of the packet was more than the default

as the reference points. On the outgoing

MRU, and so the packet was discarded. Therefore,

direction, it was only necessary to check the

engineers could not ping successfully with packet

MTU. The length of the HTTP request packet

length more than 1492 bytes from PC2.

Internet by PC2 was 1492 bytes (L2 packet

Solution

length = L3 packet length + 18). Even if the

Engineers modified the MTU on GER to 1500

packet was encapsulated with two layers of

bytes. The fault still persisted. As GER of the current

MPLS labels, the total length was still less

version did not support MRU modification, engineers

than 1500 bytes. Therefore, there was no

upgraded the device to V2.6.03.B.52.P10 and then

problem from PC to the Internet.

modified the MRU to 1608 bytes. The fault was solved.

2. At the T64E side: On the incoming


direction, it was only necessary to check

Maintenance Experience

one MPLS label, that is, the VPN label. As a result, the

1. On the network, take PC1 and PC2

or TCP acknowledge packet sent to the

24

GER enabled penultimate hop popping by default,


so the packet forwarded to GER by T1200 contained

Summary

the MRU. Generally, the length of an HTTP

For cases that jumbo packets cannot be pinged

packet sent from the Internet to PC1 was

successfully, we usually take the MTU configuration

1500 bytes or even larger. After the packet

into account first. We usually ignore the MRU,

was encapsulated with two layers of MPLS

because the default MRUs on many devices are large

labels, the total length of the packet was

and these devices do not support MRU modification.

consequentially more than the default MRU

The default MRU values on devices of V2821 are

(1500 bytes). The MRU on the interface

listed below:

of T64E was 1700 bytes. Therefore, on

Ethernet interface (fei, gei and xgei): 7600 bytes

the incoming direction, the packets with

Oc interface (pos48 and pos12): 7600 bytes

the length less than 1700 bytes could be

Oc interface (pos3): 1700 bytes

forwarded.

Atm interface (atm622): 7600 bytes

If there were jumbo packets with

Atm interface (atm155): 1700 bytes

the length more than 1700 bytes, it was

Oc interface (cpos3): 1700 bytes

necessary to take fragment into account.

Ce1 interface: 7600 bytes

www.zte.com.cn

VPLS Service Failure on T160G


Yang Fei / ZTE Corporation

Key words: T160G, VPLS, NP pinch plate

Network Topology
The network topology is shown in Figure 1.

xconnect vfi test

interface fei_1/1

switchport access vlan 200

interface fei_1/2

switchport access vlan 100

mpls ip
mpls

ldp

61.139.36.33

target-session

ip route 0.0.0.0 0.0.0.0


59.43.1.2

The configuration related to VPLS on


T160G-2 is shown below:

Figure 1. Network Topology

Two T160G switches interconnected with each


other through VLAN 100 on interfaces fei_1/2.

PCs were connected to the switches through


VLAN 200 on interfaces fei_1/1.

It was required to make the PC ping each other


successfully through VPLS function on the
switches.
The configuration related to VPLS on T160G-1

is shown below:

pwtype ethernet-vlan
peer 61.139.36.33

interface loopback 1
255.255.255.255

interface vlan 100

61.139.36.34

ip address 59.43.1.1 255.255.255.252


mpls ip

interface vlan 200

pwtype ethernet-vlan
peer 61.139.36.34

interface loopback 1

ip address 61.139.36.33

255.255.255.255

interface vlan 100


ip

address

255.255.255.252

59.43.1.2

interface vlan 200

vcid 200

address

vcid 200

mpls ip

vfi test

ip

vfi test

xconnect vfi test

interface fei_1/1

switchport access vlan 200

interface fei_1/2

switchport access vlan 100

mpls ip
mpls

ldp

61.139.36.34

target-session

ip route 0.0.0.0 0.0.0.0


59.43.1.1

Data Products

25

June 2010

Issue 219

Malfunction Situation
After the configuration, when engineers set the PCs to be in the same segment 1.1.1.0/24, the
PCs could not ping each other successfully.

Malfunction Analysis

1. Engineers checked the VC states on the two switches. The VC states were up, as shown
below:
T160G-1#show mpls l2transport vc vpls vfi test
VFIName
Test

LocalCircuit
VFI

DestAddress

61.139.36.33

Test

LocalCircuit
VFI

Status

VC ID

Status

200

T160G-2#show mpls l2transport vc vpls vfi test


VFIName

VC ID

DestAddress

61.139.36.34

200

up

up

2. Engineers checked the distribution of out-labels on the two switches. The labels were
proper, as shown below:
T160G-1#show mpls forwarding-table
Mpls Ldp Forwarding-table:
InLabel
53

OutLabel

Dest

101712

Pfxlen

61.139.36.33

32

Interface

NextHop

fei_1/2

59.43.1.2

T160G-2#show mpls forwarding-table


Mpls Ldp Forwarding-table:
InLabel
101711

OutLabel
54

Dest

61.139.36.34

Pfxlen
32

Interface
fei_1/2

NextHop

59.43.1.1

3. Engineers checked the distribution of in-labels on the two switches. The labels were proper.
4. Engineers checked the MAC learning on the two switches. They found that the switches
could not learn the MAC addresses of the PCs, as shown below:
T160G-1#show mac vfi test
MAC

peer-address

outIntLab

T160G-2#show mac vfi test


MAC

peer-address

outIntLab

outExtLab

outInterface

type

outExtLab

outInterface

type

5. Engineers thought that this fault was caused by the PCs, network cables or the ports on
the switches. They replaced the PCs, network cables and the ports on the switches. The fault
persisted.
6. Engineers entered diagnosis mode to check the micro code versions of the NP pinch plates
on the two switches. They found that the micro code versions of the NP pinch plates on T160G
switches were the default proper version, as shown below. However, to support VPLS function, G
series devices should use NP pinch plates with micro code of vpn version.
T160G-1 (diag)#npc 1 np chip-info microcode-version

26

Maintenance Experience

www.zte.com.cn

Mcode normal version is running!


T160G-2(diag)#npc 1 np chip-info
microcode-version

Mcode normal version is running!

Solution

T160G-2(config)#net-card vpn
T160G-2#reload slot 1

Summary
To support VPLS service, G series
switches need NP pinch plates. There is

Engineers modified the version of NP pinch

one more condition, that is, the micro code

plates to vpn on the switches and then reset

version of the NP pinch plates must be

the two boards. MAC addresses could be

vpn. By default, the micro code version of

learnt properly, and the PCs could ping each

the NP pinch plates is proper. Therefore,

other successfully. The fault was solved. The

it is necessary to modify the micro code

configuration commands are shown below:

version and reset the boards plugged in

T160G-1(config)#net-card vpn
T160G-1#reload slot 1

the slots corresponding to the NP pinch


plates.

Experiences About Flow Mirroring


Configuration on ZXUAS 10800
Wang Huali / ZTE Corporation
Key words: ZXUAS 10800, mirroring, PPPoE, IPoE, subscriber

Malfunction Situation
As shown in Figure 4, 9806E router connected
to the port Ethernet 1/2 of UAS 10800. A
modem connected to the 9806E router. Local
authentication was configured on UAS 10800. The
modem obtained IP address, DNS server address

successfully. However, the PC did not


capture any PPPoE dialing packets.

Malfunction Analysis
The configuration of traffic mirroring on
UAS 10800 is shown below.

and other information through PPPoE dialing to get


online.
It was required to mirror the traffic on port 1/2
to port 1/10 and use a PC to capture packets for
analysis. After traffic mirroring was configured,
the modem could dial up to obtain IP address

Figure 1. Network Topology

Data Products

27

June 2010

Issue 219

context pppoe9806e
domain 9806e.com
forward policy mirror2
/*create a forwarding policy
mirror2*/
mirror destination
mirror_dest2 all
/*mirror all traffic to the
logical target mirror_dest2*/
interface mirrorport
ip address 88.1.1.1/24
!
interface pppoe_9806e
multibind
ip address 10.12.102.1/24
ip pool 10.12.102.0/24
!
subscriber default
ip address pool
dns primary 172.24.242.242
!
subscriber name tw
password tw
port ethernet 1/2
description to_9806e
no shutdown
encapsulation pppoe
bind authentication chap pap
maximum 500
forward policy mirror2 in
/*incoming traffic mirroring*/
forward policy mirror2 out
/*outgoing traffic mirroring*/
port ethernet 1/10
medium speed 100 duplex half
no shutdown
bind interface mirrorport
pppoe9806e
/*bind to an interface*/
forward output mirror_dest2
/*the destination mirroring
port working as the egress of
mirror_dest2*/

[local]Redback(config)#show
detail
ethernet 1/10 state is Up
Description
Line state
Admin state
Link Dampening
disabled
Undampened line state
Dampening Count
Encapsulation
ethernet
MTU size
Bytes
NAS Port Type
Speed
Mbps
Duplex mode
MAC address
00:30:88:01:79:05
Media type
100Base-TX
Loopback
Active Alarms

port 1/10
:
: Up
: Up
: Up
: 0

: 1500
:

: 100

: half

:
:

: off
: NONE

Engineer executed show port counter 1/10


command. They found that there were several
thousand broadcast packets. This meant that the
traffic was not mirrored to port 1/10.

Solution
Engineers modified the configuration, as shown
below:
subscriber default
ip address pool
dns primary 172.24.242.242
forward policy mirror2 in
forward policy mirror2 out
Tr a f f i c m i r r o r i n g s u c c e e d e d a f t e r t h e
configuration was modified. The PC could capture
the PPPoE dialing packets.
The malfunction reason was that PPPoE

1. According to the situation, traffic


mirroring did not succeed. Engineers

PPoE, it was necessary to apply the forwarding

checked the configuration commands

policy to subscriber.

commands were correct.

Maintenance Experience

There was no exception, as shown below.

encapsulation was enabled on port 1/2, and for

according to the configuration manual. The

28

2. Engineer checked the state of port 1/10.

Traffic mirroring function is only valid to IP


traffic.

Das könnte Ihnen auch gefallen