Poor Switch Performance

Most performance issues on switches are related to one of three errors:

1. Cabling and port problems (layer 1)
2. Duplex mismatches between switch ports and an attached device
3. TCAM issues.

Physical layer Troubleshooting Commands

show interface

show interface counters

show interface counters errors

Look for the following errors:

FCS-Err
Usually a cabling issue.

Xmit-Err
The transmission buffers are full.  This is sometimes seen when switching from a fast link to a slower one.

Undersize, Giants
The transmitting NIC may have problems.

Single-Col, Multi-Col, Late-Col, Excess-Col
All of these are collision types, which can point to a duplex mismatch.  Cisco recommends setting all interfaces, switch and server, to auto.
 

Spanning Tree

Spanning Tree Protocol is a loop prevention mechanism to allow redundant Ethernet network connections.  Here is an important summary of how each switch determines Spanning Tree port roles:

1. Each switch periodically transmits BPDUs that include its bridge ID, current root bridge, and cost to that root bridge.  Additionally, each switch starts assuming it is the root bridge.

2. If a switch receives a BPDU from another switch with a different root, it does a comparison.  If the BPDU has a lower advertised root, the switch changes its root to match and recalculates the cost to the new root.  The port that received the BPDU is now the root port – all others become designated ports.

3. If a switch receives two BPDUs with the same root, it then compares costs and uses the port with the lowest cost.  The port with the higher cost is blocked – also called a non-designated port.

To quickly review STP costs, below is a list of link costs based on interface speed.

After the whole process, there will be only one root bridge – with each non-root switch having only one root port.

To see the status of spanning tree, do a show spanning-tree vlan vlan-id.

To view sent/received BPDU information for a switch, do a show spanning-tree interface interface detail.

Broadcast Storms

Broadcasts storms can occur due to Spanning Tree misconfigurations and/or rogue switches being added which closes a loop.  Regardless, a broadcast storm will be obvious when the switch slows way down, becoming unresponsive, and all the links light up solid green.

The CLI may be very slow to respond if you still have remote access to it, so often times to fastest way to fix the problem is to physically begin pulling redundant links.
 

Troubleshooting EtherChannels

EtherChannel issues usually fall into one of three categories:

1. Every port participating in an EtherChannel must have identical speed, duplex, access or trunk settings.  If an EtherChannel isn’t forming, check each port configuration.

2. Both sides of the EtherChannel must be configured as a bundle directly or be using a link aggregation protocol (LACP or PAgP).  If one side is configured as an EtherChannel and the other side is not, look for error-disabled EtherChannel ports on the EtherChannel-enabled switch.

3. If traffic is only flowing over a single link in a bundle, it is likely that the hash algorithm should be adjusted to use different seed values.  Also note that link bundles should be used in even numbered pairs like 2, 4, 8, etc.
 

VLANs

When troubleshooting issues that you suspect are related to VLAN logic, you should first make sure you have tested for physical layer issues like bad cabling, a power failure, or bad switch ports.  Also, check that you are not dealing with an issue with the switch itself – things like software bugs, loops, or ARP problems.

VLAN issues usually come in the form of misconfigured VLANs, improper VTP mode, trunk issues, or native VLAN mismatches. 

Switch Tables

It is important that you understand what show commands display information on what switch tables.  These will come in handy when you are isolating a switching issue.

MAC Address Table (MAC-to-port mapping)

show mac-address

VLAN Assignments (VLAN-to-port mapping)

show vlan

Trunk Assignments

show interface switchport

show interface switchport trunk

show etherchannel 

Troubleshooting Inter-VLAN Routing

Routing between VLANs can be done on either a router, or a layer 3 switch – but the data plane is different depending on the platform you are using.

Either way, show ip cef displays the CEF forwarding table and show adjacency will show you the layer 2 headers used in forwarding.

Keep in mind that routers always use layer 3 information to pass traffic between ports.  Switches can either use MAC address forwarding (for layer 2 forwarding), SVIs for inter-VLAN routing, or layer 3 routed ports.  The last category, routed ports do not run layer 2 protocols like Spanning Tree – very important.

Last thing to remember – SVIs will only go into down state when all interfaces within that particular VLAN are down.

HSRP, VRRP, & GLBP

First hop redundancy protocols allow a layer 2 segment to have two gateway routers for redundancy, while still only showing a single gateway IP and MAC address.

The three FHRPs Cisco supports are HSRP, VRRP, and GLBP.

HSRP is one of the original FHRPs that was developed by Cisco and is proprietary.  One router is active and another is a backup (using HSRP keepalives to maintain connectivity).  HSRP is extremely popular and you should make sure to understand how it works for the TSHOOT exam.  Check out the High-Availability page to learn more.

VRRP is another gateway redundancy protocol that is an open standard and very similar to HSRP.

GLBP is Cisco proprietary; its primary advantage is its ability to automatically load balance between gateway routers.

HSRP

HSRP is the primary FHRP covered on the TSHOOT exam, so let’s go through the basics one more time.

HSRP is configured using the standby command under interface configuration mode.  Routers in the same HSRP group share a common MAC and virtual IP address.  The standby configuration statements define the HSRP group as well as the virtual IP in use.

Each HSRP-enabled router has a default HSRP priority of 100 (remember, highest wins).  If another router joins the group with a higher priority it will still not become the active router unless the preempt command is applied.

An example HSRP configuration could look something like:

Router(config)# interface gig1/1
Router(config-if)# ip address 192.168.1.2
Router(config-if)# standby 4 ip 192.168.1.1
Router(config-if)# standby 4 priority 200
Router(config-if)# standby 4 preempt

To show the current HSRP status, issue either show standby or show standby brief depending on the level of detail you require.