Document ID: 71990
Contents
Introduction
Prerequisites
Requirements
Components Used
Conventions
Background Information
Troubleshoot Common High CPU Utilization Problems
High CPU Utilization After Enabling GRE Tunnels
High CPU Utilization When Pinging Own Interface
High CPU Utilization Due to VUR_MGR bg Process
High CPU Utilization Due to IP Input Process
High CPU Utilization Due to GigaStack GBIC Modules
High CPU Utilization Due to TTY Background Process
High CPU Utilization Due to SNAP Encapsulation of IPv4 Packets
High CPU Utilization Due to IP Redirects
High CPU Utilization Due to Broadcast Storm
NetPro Discussion Forums - Featured Conversations
Related Information
Introduction
This document describes causes of high CPU utilization on the Cisco Catalyst 3550 Series Switches. This document also lists common network or configuration scenarios that can cause high CPU utilization on the Catalyst 3550 Series Switches.
Cisco Catalyst switches use the show processes cpu command in order to identify the causes of high CPU utilization. The show processes cpu command shows CPU utilization averaged over the past five seconds, one minute, and five minutes. CPU utilization numbers do not provide a true linear indication of the utilization with respect to the offered load. These are some of the major reasons:
-
In a real world network, the CPU has to handle various system maintenance functions, such as network management.
-
The CPU has to process periodic and event-triggered routing updates.
-
There are other internal system overhead operations, such as polling for resource availability, that are not proportional to traffic load.
Prerequisites
Requirements
There are no specific requirements for this document.
Components Used
The information in this document is based on Catalyst 3550 Series Switches.
The information in this document was created from the devices in a specific lab environment. All of the devices used in this document started with a cleared (default) configuration. If your network is live, make sure that you understand the potential impact of any command.
Conventions
Refer to Cisco Technical Tips Conventions for more information on document conventions.
Background Information
Before you look at the CPU packet-handling architecture and troubleshoot high CPU utilization, you must understand the different ways in which hardware-based forwarding switches and Cisco IOS® Software-based routers use the CPU. The common misconception is that high CPU utilization indicates the depletion of resources on a device and the threat of a crash. A capacity issue is one of the symptoms of high CPU utilization on Cisco IOS routers. However, a capacity issue is almost never a symptom of high CPU utilization with hardware-based forwarding switches.
The CPU utilization of 20% to 50% is normal on a Catalyst 3550 Switch, even under minimal load. CPU utilization does not reflect the total number of packets being switched or the total load on the switch. The CPU is responsible for the process of IP traffic (broadcast, telnet, SNMP) on the management VLAN, for the process of control packets Spanning-Tree Protocol (STP), Cisco Discovery Protocol (CDP), DDSN Transfer Protocol (DTP), Port Aggregation Protocol (PAgP), Link Aggregation Control Protocol (LACP), Unidirectional Link Detection (UDLD), address learning, routing protocols, port status and LED operations. If the CPU utilization is extremely high (around 90% - 99%), this does not directly affect the switching of data. However, high CPU utilization might start to affect protocols such as STPs. The CPU on the Catalyst 3550 is used for management purposes only. The CPU is not used to forward packets. This is handled by ASICs. The increase in the CPU utilization does not affect traffic forwarding.
The first step to troubleshoot the high CPU utilization is to check the Cisco IOS version release notes of your Catalyst 3550 Switch for the possible known IOS bug. This way you can eliminate the IOS bug from your troubleshooting steps. Refer to Cisco Catalyst 3550 Series Switches Release Notes for a description of new features, system requirements, limitations, restrictions, caveats, and troubleshooting information for a particular software release for Catalyst 3550 Switches.
Troubleshoot Common High CPU Utilization Problems
This section provides information about some of the common high CPU utilization problems on the Catalyst 3550 Switch.
High CPU Utilization After Enabling GRE Tunnels
Generic Routing Encapsulation (GRE) tunnels are not supported on the Cisco Catalyst 3550 Switch. Even though the CLI commands are there to configure the GRE, it is not officially supported. Refer to the Unsupported VPN Configuration Commands section of Unsupported CLI Commands for Catalyst 3550 for this information. The reason for this is that the Cisco Catalyst 3550 Switch uses hardware-based Cisco Express Forwarding (CEF) switching. There is no method to CEF-switch GRE packets. GRE packets must be encapsulated by the software. The hardware does not have the capability to encapsulate the packets. Consequently, this traffic is processed or software switched. The process or software switched traffic can quickly cause the CPU to spike.
High CPU Utilization When Pinging Own Interface
An extended ping from one interface to another interface on the same
switch can cause high CPU utilization. This can occur when a large number of
ping packets are sent and received. This is an expected behavior. The
workaround is to not perform a ping from one interface to another on the same
switch. Refer to Cisco bug ID
CSCea19301
(
registered customers only)
for more
information.
High CPU Utilization Due to VUR_MGR bg Process
The VUR_MGR bg process is the Vegas Unicast Routing Manager process which is a platform specific module that interfaces with IOS. This process implements the hardware independent fuctionality required in the platform for unicast routing. Each time an Address Resolution Protocol (ARP) is resolved for a destination, the corresponding entry needs to be programmed in hardware.
The VUR_MGR bg process is responsible for unicast routing and is high if the switch is learning routing information. It is also high if you see frequent routing changes. Issue the clear ip route command in exec mode to clear the condition. However, this does not prevent the condition to recur.
Cat-3550#show processes cpu CPU utilization for five seconds: 99%/0%; one minute: 99%; five minutes: 99% PID Runtime(ms) Invoked uSecs 5Sec 1Min 5Min TTY Process 1 4 1103385 0 0.00% 0.00% 0.00% 0 Load Meter 2 52592 5709333 9 0.00% 0.00% 0.00% 0 Spanning Tree 3 1897604 550508 3447 0.00% 0.01% 0.00% 0 Check heaps 4 0 1 0 0.00% 0.00% 0.00% 0 Chunk Manager 5 12268 59126 207 0.00% 0.00% 0.00% 0 Pool Manager 6 0 2 0 0.00% 0.00% 0.00% 0 Timers 7 12 2 6000 0.00% 0.00% 0.00% 0 Entity MIB API 8 1244 1101072 1 0.00% 0.00% 0.00% 0 HC Counter Timer 9 3340 99041 33 0.00% 0.00% 0.00% 0 ARP Input 10 0 2 0 0.00% 0.00% 0.00% 0 Net Input 11 0 1 0 0.00% 0.00% 0.00% 0 Critical Bkgnd 12 13172 591932 22 0.00% 0.00% 0.00% 0 Net Background 13 0 61 0 0.00% 0.00% 0.00% 0 Logger 14 7644 5465294 1 0.00% 0.00% 0.00% 0 TTY Background 15 9608 5465305 1 0.00% 0.00% 0.00% 0 Per-Second Jobs 16 0 2 0 0.00% 0.00% 0.00% 0 Vegas Storm Cont 17 88 126356607 0 0.00% 0.00% 0.00% 0 Vegas LED Proces 18 0 1 0 0.00% 0.00% 0.00% 0 SCQ_PROCESS 19 20 238 84 0.00% 0.00% 0.00% 0 RAM Access 20 84020 71566887 1 0.00% 0.00% 0.00% 0 SW Frame Ager 21 4 1103386 0 0.00% 0.00% 0.00% 0 Compute load avg 22 368684 91998 4007 0.00% 0.00% 0.00% 0 Per-minute Jobs 23 6288 10814019 0 0.00% 0.00% 0.00% 0 L2TM Process 24 132772 15688292 8 0.00% 0.00% 0.00% 0 Vegas Statistics 25 10056 10814019 0 0.00% 0.00% 0.00% 0 L3TM 26 40 70646868 0 0.00% 0.00% 0.00% 0 HMATM Learn proc 27 4808 5465299 0 0.00% 0.00% 0.00% 0 HMATM Age proces 28 8 42 190 0.00% 0.00% 0.00% 0 VL2MM 29 12896 5465325 2 0.00% 0.00% 0.00% 0 L3MD 30 306288 13114927 23 0.00% 0.00% 0.00% 0 L3MD_STAT 31 0 1 0 0.00% 0.00% 0.00% 0 Vegas Bridging 32 168 68417996 0 0.00% 0.00% 0.00% 0 VegasPM 33 506543096 3512719 144203 99.67% 99.79% 99.95% 0 VUR_MGR bg proce !--- Output suppressed.
Refer to Cisco bug ID
CSCdx31480
(
registered customers only)
for more information.
This bug is fixed in Cisco IOS Software Release 12.1(11)EA1 and later.
High CPU Utilization Due to IP Input Process
The Cisco IOS software process called IP input takes care of process-switching IP packets. If the IP input process uses unusually high CPU resources, the switch is process-switching a lot of IP traffic. Refer to Troubleshooting High CPU Utilization in IP Input Process for information on how to troubleshoot high CPU utilization due to the IP input process.
High CPU Utilization Due to GigaStack GBIC Modules
When you insert a GigaStack GBIC in a GBIC module slot, the CPU utilization increases by six percent. This increase occurs for each GigaStack GBIC added to the switch. The VegasPM process in the show processes cpu command output shows this CPU utilization. The VegasPM process manages the Gigastack GBIC operation on the switch.
Cat-3550#show processes cpu CPU utilization for five seconds: 73%/0%; one minute: 67%; five minutes: 66% PID Runtime(ms) Invoked uSecs 5Sec 1Min 5Min TTY Process 1 0 1 0 0.00% 0.00% 0.00% 0 Chunk Manager 2 9008 2476157 3 0.00% 0.00% 0.00% 0 Load Meter 3 700 722 969 0.00% 0.00% 0.00% 0 SpanTree Helper 4 3607596 1260465 2862 0.08% 0.02% 0.00% 0 Check heaps 5 10197340 22011722 463 0.00% 0.01% 0.06% 0 Pool Manager 6 0 2 0 0.00% 0.00% 0.00% 0 Timers 7 32 20 1600 0.00% 0.00% 0.00% 0 Entity MIB API 8 55160 2475564 22 0.00% 0.00% 0.00% 0 HC Counter Timer 9 8469276 82844936 102 0.40% 0.10% 0.08% 0 ARP Input 10 24 1592 15 0.00% 0.00% 0.00% 0 Net Input 11 0 1 0 0.00% 0.00% 0.00% 0 Critical Bkgnd 12 338884 6258667 54 0.00% 0.00% 0.00% 0 Net Background 13 64 1035 61 0.00% 0.00% 0.00% 0 Logger 14 122036 12262545 9 0.00% 0.00% 0.00% 0 TTY Background 15 623312 12262563 50 0.00% 0.00% 0.00% 0 Per-Second Jobs 16 20084 2476157 8 0.00% 0.00% 0.00% 0 Compute load avg 17 1048664 206875 5069 0.00% 0.00% 0.00% 0 Per-minute Jobs 18 0 2 0 0.00% 0.00% 0.00% 0 Vegas Storm Cont 19 8455544 279108955 30 0.08% 0.02% 0.00% 0 Vegas LED Proces 20 0 1 0 0.00% 0.00% 0.00% 0 SCQ_PROCESS 21 1800 2468 729 0.00% 0.00% 0.00% 0 RAM Access 22 8175752 229227255 35 0.00% 0.04% 0.04% 0 SW Frame Ager 23 456 1104 413 0.00% 0.00% 0.00% 0 VLAN Info Update 24 205420 23909802 8 0.00% 0.00% 0.00% 0 L2TM Process 25 44726932 24487858 1826 0.24% 0.37% 0.38% 0 Vegas Statistics 26 617288 23909801 25 0.00% 0.01% 0.00% 0 L3TM 27 4122052 141972837 29 0.00% 0.00% 0.00% 0 HMATM Learn proc 28 197324 12262549 16 0.00% 0.00% 0.00% 0 HMATM Age proces 29 495616 2291212 216 0.00% 0.01% 0.00% 0 VL2MM 30 2175000 18368663 118 0.00% 0.00% 0.00% 0 L3MD 31 142702548 78397696 1820 1.30% 1.13% 1.11% 0 L3MD_STAT 32 0 1 0 0.00% 0.00% 0.00% 0 Vegas Bridging 33 3607116392 397152327 9082 61.46% 63.81% 63.05% 0 VegasPM 34 121392 6295553 19 0.00% 0.00% 0.00% 0 VUR_MGR bg proce !--- Output suppressed.
As a workaround, if the network design permits, use other types of
GBICs such as fiber. These do not cause additional CPU utilization. Refer to
Cisco bug ID
CSCdx90515
(
registered customers only)
for more
information.
High CPU Utilization Due to TTY Background Process
The TTY Background process is a generic process used by all terminal lines (console, aux, async, and so on). Normally there should not be any impact on the performance of the switch, because this process has a lower priority compared to the other processes that need to be scheduled by the Cisco IOS software.
If this process takes high CPU utilization, check whether logging
synchronous is configured under line con 0. Refer to Cisco bug ID
CSCdy01705
(
registered customers only)
for more
information.
High CPU Utilization Due to SNAP Encapsulation of IPv4 Packets
On the Catalyst 3550 Switch, Layer 3 forwarding of IPv4 in the Subnetwork Access Protocol (SNAP) can only be done in the software. SNAP-encapsulated IPv4 packets that are directed to the router MAC address or the Hot Standby Router Protocol (HSRP) group MAC address (if this is the active router in the HSRP group) are forwarded to the switch CPU. This action can potentially cause high CPU utilization levels.
Packets received from media types that require SNAP encapsulation of IPv4 packets require the switch to forward SNAP-encapsulated packets. In general, Layer 2 forwarding of IPv4 in SNAP encapsulation takes place in the hardware, unless a VLAN map or port Access Control List (ACL) contains an IP ACL. However, this cannot take place on the Cisco Catalyst 3550 Switch.
This is a hardware limitation, and there is no workaround. Refer to
Cisco bug ID
CSCed59864
(
registered customers only)
for more
information.
High CPU Utilization Due to IP Redirects
ICMP redirect messages are used by routers and switches to notify the hosts on the data link that a better route is available for a particular destination. By default, Cisco routers and switches send ICMP redirects.
You can expect the sourcing device to act on the ICMP redirect that the Catalyst 3550 sends, and to change the next hop for the destination. However, not all devices respond to an ICMP redirect. If the device does not respond, the Catalyst 3550 must send redirects for every packet that the switch receives from the sending device. These redirects can consume a great deal of CPU resources. The high CPU utilization is caused due to the high amount of ICMP redirect traffic that hits the CPU. The workaround is to disable ICMP redirects with the no ip redirects interface mode command.
This scenario can also occur when you have configured secondary IP addresses. When you enable the secondary IP addresses, the IP redirect is automatically disabled. Make sure you do not manually enable the IP redirects when you have configured secondary IP addresses.
High CPU Utilization Due to Broadcast Storm
The CPU utilization of a switch can go up if a large number of broadcast packets is received.
3550-1>show int gi0/1
GigabitEthernet0/1 is up, line protocol is up (connected)
Hardware is Gigabit Ethernet, address is 0014.698a.bb31 (bia
0014.698a.bb31)
MTU 1500 bytes, BW 1000000 Kbit, DLY 10 usec,
reliability 255/255, txload 1/255, rxload 1/255
Encapsulation ARPA, loopback not set
Keepalive set (10 sec)
!--- Part of the output elided.
5 minute input rate 3660000 bits/sec, 382 packets/sec
5 minute output rate 7413000 bits/sec, 2119 packets/sec
25774872 packets input, 1350686943 bytes, 0 no buffer
Received 25774872 broadcasts (0 multicast)
0 runts, 0 giants, 0 throttles
0 input errors, 0 CRC, 0 frame, 0 overrun, 0 ignored
!--- Rest of the output elided.
If these broadcast storms are frequent, then you might have to look into the design of the network. If the broadcast storms are occasional, you can configure the Storm Control feature in order to equip the device against the storm. Refer to Configuring Port-Based Traffic Control for more information.
NetPro Discussion Forums - Featured Conversations
| NetPro Discussion Forums - Featured Conversations for LAN |
| Network Infrastructure: LAN Routing and Switching |
| Network Infrastructure: Getting Started with LANs |
Related Information
- The show processes Command
- Troubleshooting High CPU Utilization due to Processes
- High CPU Utilization on Catalyst 2900XL/3500XL Switches
- Catalyst 3750 Series Switches High CPU Utilization Troubleshooting
- Catalyst 6500/6000 Switch High CPU Utilization
- Troubleshooting High CPU Utilization on Cisco Routers
- LAN Product Support Pages
- LAN Switching Support Page
- Technical Support & Documentation - Cisco Systems
| Updated: Nov 28, 2006 | Document ID: 71990 |