LPU board High cpu usage by ARP Packets Accumulated on 3G network.

1.topology

N/A

2.version:

V600R001C01SPC600

3.problem detail:

Customer face the issue found one LPU board CPU-usage 99%, the 3G services effect.

dis device slot 3

Slot 3's detail information:

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Description: Line Processing Unit F-21-A

Board status: Normal

Register: Registered

Uptime: 2011/08/01 08:05:50

CPU Utilization(%): 99%

Mem Usage(%): 58%

display alarm all hist

290 Error 11-07-29 00:52:05 LPU 3 is failed, NP-3 PCS Status of PI

C0 is abnormal

291 Error 11-07-29 00:52:36 LPU 3 is failed, NP-3 PCS Status of PI

C0 is abnormal, Resume

1. first check the configuration on interface.

interface GigabitEthernet3/1/1.10

 description NE-MUX-NODE-B-vlan-2000-to-2111

 control-vid 10 dot1q-termination

 dot1q termination vid 10

 dot1q termination vid 2000 to 2111

 dot1q vrrp vid 10

 ip binding vpn-instance IUB

 ip address 10.72.149.2 255.255.255.128

 vrrp vrid 10 virtual-ip 10.72.149.1

 arp broadcast enable

2.check the cpu-usage on LPU board.found the ARP process have 74%.

dis cpu-usage slot 3

CPU Usage Stat. Cycle: 60 (Second)

CPU Usage : 99% Max: 99%

CPU Usage Stat. Time : 2011-08-01 15:49:19

CPU utilization for five seconds: 99%: one minute: 99%: five minutes: 99%.

...

ARP 74% 0/6e37bd6a ARP

...

3.after check the arp message queue,found this queue have congestion.that means ARP requrest process use the high usage.

[NE40E-PUN-1-hidecmd]display  message-queue  slot 3

*******************************************************

 Max Queue Count     = 256

 Current Queue Count = 163

 Total Created Count = 163

 Default Queue Size  = 0x300

-------------------------------------------------------

  QID   Mode       TotLen  CurLen  MaxSize  Name     

-------------------------------------------------------

  1     FIFO SYN   200     3       16       PATQ      

......

  143   FIFO ASY   10000   8825    16       Q2PK  

......

also the Que0 currlen the rate of growth had sustained,that means the CPU usage will come to high.

[NE40E-PUN-1-hidecmd]display umsg queue utask-id 53 slot-id 3

-----------------------------------------------------

QueName         MaxLen         CurrLen         Status

-----------------------------------------------------

Que0            100          13561              UnLock

Que1            100              0              UnLock

[NE40E-PUN-1-hidecmd]display umsg queue utask-id 53 slot-id 3

-----------------------------------------------------

QueName         MaxLen         CurrLen         Status

-----------------------------------------------------

Que0            100          15860              UnLock

Que1            100              0              UnLock

When use the Dot1q termination sub-interface receives a large number of unknown unicast packets, ARP Miss packets are triggered. Each ARP Miss packet causes the system to write a UMSG message to the Utask task. The system schedules UMSG messages to send ARP requests. As the LPU keeps receiving unknown unicast packets, ARP Miss packets are generated frequently. The system processes UMSG messages at a speed lower than the speed at which ARP Miss packets trigger ARP Requests. As a result, memory usage of the LPU increases. After the memory usage exceeds the pre-configured threshold, the LPU is reset.

Suggestion update the new patch. patch name V600R001C00SPC026.