Figure 5-1 shows the networking of an enterprise. The CPU usage of an S5700 (S11) frequently exceeds 90%.
Figure 5-1 Receiving many STP TC packets causes a high CPU usage on an S5700
Run the display cpu-usage command to check CPU information on the S5700. The CPU usage of the S5700 has reached 97%.
<S5700> display cpu-usage
CPU Usage Stat. Cycle: 60 (Second)
CPU Usage : 18% Max: 97%
CPU Usage Stat. Time : 2014-10-07 11:19:29
CPU utilization for five seconds: 18%: one minute: 18%: five minutes: 18%
Max CPU Usage Stat. Time : 2014-09-11 16:37:54.
View log information on the S5700. The log shows that a large number of TC packets have been processed.
Step 1 Related information was not collected when the fault occurred; therefore, it is unknown that which process triggers high CPU usage. There is a possibility that the FTS task triggers high CPU usage when processing many TC packets. The switch continuously generates TC packet logs. Check whether the TC packets are generated by the switch or received from other devices.
Run the display stp tc-bpdu statistics command on the S5700. The command output shows that the number of TC packets received by GigabitEthernet0/0/52, which is connected to SwitchA, keeps increasing, and the TC packets are forwarded to other access switches. Therefore, the TC packets are not generated by the S5700.
Step 3 The command output on S4 shows that the number of outgoing TC packets keeps increasing; therefore, the TC packets are generated by S4. Run the display stp topology-change command on S4 to view TC packet information. The following information shows that the status of GigabitEthernet0/0/51 alternates between Blocked to Forwarding. When the port status changes to detected, topology changes.
<S4> display stp topology-change
CIST topology change information
Number of topology changes :8233
Time since last topology change :0 days 0h:0m:26s
Topology change initiator(detected) :GigabitEthernet0/0/51
Number of generated topologychange traps : 9852
Number of suppressed topologychange traps: 13
Step 4 Run the display interface brief command to view interface information on access devices. GigabitEthernet0/0/51 has received a large number of error packets. Wait for a period of time, and check the interface information again. GigabitEthernet0/0/51 still received a large number of error packets. This indicates that the fiber connected to this port is abnormal. After the fiber is replaced, the fault is rectified.
<S4> display interface brief
*down: administratively down
(E): E-Trunk down
(b): BFD down
(e): ETHOAM down
(dl): DLDP down
(d): Dampening Suppressed
InUti/OutUti: input utility/output utility
Interface PHY Protocol InUti OutUti inErrors outErrors
GigabitEthernet0/0/51 up up 0.01% 0.02% 38068638 0
On an STP network, the quality of links between the devices participating in STP calculation degrades, causing frequent topology convergence and a large number of TC packets. Therefore, the CPU usage of the devices receiving the TC packets increase, affecting services.
Find out the reason why the quality of links between S3 and S4 degrades.
Run the stp tc-protection command globally on the core devices participating in STP calculation. After the command is executed, a device updates entries at most once every two seconds when the device frequently receives TC packets. This configuration prevents high CPU usage caused by frequent updates of MAC address entries and ARP entries.