Guest

Cisco CSS 11500 Series Content Services Switches

Content Services Switch FAQ

Cisco - Content Services Switch FAQ

Document ID: 25901

Updated: Apr 19, 2006

   Print

Contents

Introduction

This document addresses the most frequently asked questions (FAQ) about the Cisco Content Services Switch (CSS).

Refer to Cisco Technical Tips Conventions for more information on document conventions.

Q. Where can I find the MIBs for the CSS?

A. The MIBs are already on the CSS. You can consider the CSS an agent in the Simple Network Management Protocol (SNMP) network scheme. All you need to do is configure the SNMP parameters on the CSS. Refer to the document Configuring Simple Network Management Protocol (SNMP) for more information.

Q. What is the maximum number of scripted keepalives that the CSS supports?

A. The maximum number of scripted keepalives that CSS supports is 255. Refer to the New Features in Software Version 5.00 section of the Release Note for the Cisco 11000 Series Content Services Switch.

Q. How can I clear or remove core files?

A. Issue the clear core command. The command is available in CSS software version 5.00 and later, in debug mode. The syntax is:

css150(debug)#clear core filename CR

Q. Where can I find interpretations of log messages?

A. For interpretations of log messages, refer to the document Log Messages.

Q. Is there a command that controls how often peers send load reports to each other?

A. You can use the dns-peer interval command. There are also additional commands that you can configure locally in order to achieve a quicker measure of the local load:

  • ageout-timer—Sets the time (in seconds) of the ageout of stale load information.

  • teardown-timer—Sets the maximum time period (in seconds) that the system waits to send a teardown report.

Q. Do license keys change with code versions?

A. No, license keys do not change with code versions.

Q. I lost my license key. What do I do?

A. Send an email with the serial number of your CSS to licensing@cisco.com. The version command displays the feature pack, but not the license key.

Q. What is the default time for the retention of an entry in a sticky table?

A. Unless you use the command sticky-inact-timeout, there is no default time. The sticky table is kept on a FIFO basis (32,000 or 128,000 entries, according to the device type and memory available), or until the reboot of the CSS.

Q. How do I configure sticky mask in order to cover requests from a mega-proxy like America Online (AOL)?

A. If an application requires a user to be stuck for the entire life of the session, consider a Layer 3 sticky. A Layer 3 sticky sticks a user to a server on the basis of the user IP address. The CSS has a sticky table of 32,000, which means that when 32,000 simultaneous users are on the site, the table wraps and the first users become "unstuck". However, the volume of your site can be such that you have more than 32,000 users at a time. Or a large percentage of your customers can come to you through a mega-proxy. In these cases, consider either the use of a different sticky method (such as the cookie, cookieurl, or url) or an increase of your sticky mask. The default sticky mask is 255.255.255.255, which means that each entry in the sticky table is an individual IP address. Some of the mega-proxies have a situation in which one user over the life of one session uses several different IP addresses in a range of addresses. This situation causes some of the TCP connections to get stuck to one server, and can cause other connections to get stuck to a different server for the same transaction. A result can be the loss of some items from the grocery cart. If you cannot use one of the more advanced methods of sticking, use the sticky mask of 255.255.240.0 when your client base comes through one of these mega-proxies.

Q. Why is there no option for sticky when I use advanced-balance Secure Socket Layer (SSL)?

A. Advanced-balance SSL is the same as sticky SSL.

Q. What type of encryption does Content and Application Peering Protocol (CAPP) or Application Peering Protocol (APP) use?

A. By default, CAPP uses no encryption. You can configure the APP session to use Message Digest 5 (MD5). The encryption type must be the same on both peers in order for the APP session to come up.

Q. What does the "gratuitous arp" message mean?

A. When the backup switch does not detect a heartbeat from the master switch within 3 seconds, the backup switch transitions to become the master and sends a "gratuitous arp" message. The message indicates an Address Resolution Protocol (ARP) transmittal from the new master switch. The message contains the MAC address of the current master switch. The gratuitous arp is enabled by the ip gratuitous-arps command in global configuration mode. It cannot be enabled on a single interface and block it on other interfaces.

Q. How do I synchronize configurations over the CSS in failover mode?

A. In order to synchronize configurations in software version 4.0, use the commit config sync command. In order to synchronize configurations in software version 3.10 code, you must use FTP in order to move the configuration from one switch to another. In order to synchronize configurations in software versions 6.x and 7.x code, use the command commit_redundancy for active/standby or box-to-box redundancy. Or you can use the command commit_vip_redundancy for virtual IP (VIP)/interface redundancy. You can use the command show script commit_redundancy in order to view in the header of the script the available command-line options for the commit_redundancy script. The same applies to the commit_vip_redundancy command.

Q. What settings should I use in a terminal program?

A. Use these settings:

  • 9600 baud

  • 8 bits

  • No parity

  • 1 stop bit

  • No flow control

Q. Is there a way to reprogram the MAC address on a CSS?

A. Yes, there is a way.

Note: You can find the MAC address and serial number on the back of the unit.

Complete these steps in order to reprogram the serial number and MAC address. This example is for a MAC address in the CS800 chassis:

  1. Open Offline Diagnostic Monitor (ODM).
  2. In the ODM main menu, press Shift-T in order to reach the Technician menu.
  3. Choose 1 (Configure).
  4. Choose 5 (Set Manufacture Information).
  5. Choose 2 (Set Backplane Manufacture Information).
  6. Follow the prompt and enter the data that corresponds, such as the serial number and MAC address. You can find this data on the top of the CS800 chassis.
  7. Reboot the box.

Q. How do I make a permanent prompt change on the CSS?

A. Log in to the CSS box as user fred, and use your login credentials. In order to make a permanent prompt change, issue this command:

Css100#prompt Redsox 
<cr> 
Redsox#

Issue this command to save the change:

Redsox#save_profile

This command saves the user profile so that each time the user logs in, the CSS uses the same prompt. This action, similar to use of the ?.? resource files in UNIX, creates a unique profile for each user.

When you go back to the CSS and log in as admin, the prompt does not reflect these changes. The changes are user-specific, so you need to issue the prompt and save_profile commands for each user who wants to have the prompt reflect the new change.

Q. What is the difference between operational and locked Flash?

A. This example shows the different types of Flash that the show version command displays:

CSS150-2#show version
Version:               ap0401049s (4.01 Build 49) 
Flash (Locked):        3.10 Build 33

!--- This image is the original image that was installed on the CSS.
!--- The image serves as a backup in the event that the CSS is not able 
!--- to boot from the operational Flash because of an image corruption.

Flash (Operational):   5.00 Build 10-

!--- This is the image that currently runs on the CSS.
 
Type:                  PRIMARY 
Licensed Cmd Set(s):   Standard Feature Set 
                                      Enhanced Feature Set 
                                      SSH Server 

Q. Why are there different versions of Flash?

A. Locked Flash shows the version of software that was originally installed on that CSS. The version remains the same and serves only as a backup. The version in operational Flash is the version that currently runs on that CSS.

Q. Why can I not access the management port of the CSS from a remote port?

A. In all versions of Cisco WebNS that are earlier than 5.03, the management port is not a routable interface. In version 5.03, you can add a default gateway to the management port in order to make the port a routable interface.

Q. Does Cisco Technical Support support custom-script keepalives that the customer writes?

A. No, Cisco Technical Support does not support keepalive scripts that a customer writes.

Q. How do I remove core files from the CSS disk?

A. If, after you issue the show core command, you find a list of core files, you can remove the files in one of two ways:

Note: The method you use depends on the version of code.

  • CSS50-1(config)#llama                                                      
    
    !--- This command places the CSS in debug mode.
     
    CSS50-1(debug)#clear core corefilename
     

or

  • CSS50-1(config)#llama                                                       
    
    !--- This command places the CSS in debug mode.
    
    CSS50-1(debug)#dir c:/Core/?                                          
    
    !--- This command lists the names of all the core 
    !--- files in the c:/Core directory.
    
    CSS50-1(debug)#ap_file delete c:/Core/ corefilename
        
    
    !--- This command deletes the specified core file.
    
    

Q. When I authenticate to a RADIUS server with my CSS, I get the "RADIUS-4: RADIUS Authentication failed with reason code 2" error message. What does the message mean?

A. This error message indicates that the reply has reached CSS and there is a problem. A failure to set the service-type attribute to administrative on the RADIUS server can be the cause of the problem. Check the RADIUS server and verify the service-type attributes.

Q. How large is the sticky table, and what causes the removal of entries?

A. The CSS has a 32,000 or 128,000 (which depends on the model type and memory available) sticky table that contains entries for sticky source-ip and sticky Secure Socket Layer (SSL). The sticky table does not maintain sticky cookies on the CSS. The removal of entries in the sticky table on the CSS occurs in these situations:

  • By default, with a FIFO method. Entries remain in the table until the 32,000 or 128,000 buffer is full. At this time, any new entries cause the CSS to remove an entry on the basis of FIFO.

  • sticky-inact-timeout minutes. In a content rule, you can specify the inactivity timeout by which the CSS removes a sticky entry, as this example shows:

    owner arrowpoint 
    ; content L5sticky 
    ;    vip address 192.1.1.1 
    ;    add service test1 
    ;    add service test2 
    ;    protocol tcp 
    ;    port 443 
    ;    url "/*" 
    ;    advanced-balance ssl 
    ;    application ssl 
    ;    sticky-inact-timeout 9      
    
    !--- Entry removal occurs after 9 minutes.
     
    ;    active

    Note: The CSS rejects the next sticky request in a case when all these items are true:

    • The sticky-inact-timeout parameter is used.

    • The CSS has filled the 32,000 or 128,000 buffer.

    • No entries are about to timeout.

  • Content rule. With the suspension and reactivation of a content rule, the removal of sticky table entries that apply to that rule occurs.

For more information, refer to the document Configuring Sticky Parameters for Content Rules.

Q. How can I take a service out of rotation?

A. With the configuration of the content rule (Layer 3, Layer 4, or Layer 5) as a basis, the CSS behaves differently with the manual suspension of a service, which takes a server out of service. Many times, web developers need to temporarily suspend a service and make administration changes to the web pages. Because these web changes can occur during production hours, you do not want to kill connections that exist to the service or services when the manual service suspension occurs. Perform the updates to a service during the manual service suspension.

This example shows sample Layer 5, Layer 4, and Layer 3 content rules:

owner REDSOX 
    content layer5 
    vip address 200.200.200.200 
    add service test 
    add service  test1  
    add service test2 
    protocol tcp 
    port 80 
    url "/*"      

!--- This is a Layer 5 rule.

    active 
content layer4 
     vip address 200.200.200.200 
     add service test 
     add service test1 
     add service test2 
     protocol tcp  

!--- This is a Layer 4 rule.

     port 80   

!--- This is a Layer 4 rule.
   
     active 
content layer3 
     vip address 200.200.200.200  

!--- This is a Layer 3 rule.

     add service test 
     add service test1 
     add service test2 
     active 

The CSS diverts the connections that exist when the content rules are either Layer 3 or Layer 4. If the suspension of a service under a Layer 3 or Layer 4 content rule occurs, the CSS diverts any connection that exists and forwards all subsequent TCP requests to the active service under that respective content rule.

With the manual suspension of a service that resides under a Layer 5 content rule, the CSS resets any or all connections that associate with that service.

Q. Is network proximity part of the enhanced feature set?

A. Network proximity features are not part of the enhanced feature set and require an additional license. If you try to issue proximity commands on the CSS without the appropriate license, you receive this error message:

CSS50-1(config)#proximity db 0 tier1 
                             ^ 
%% Invalid License to execute command. 
This command belongs to the Proximity Database.  Refer 
to the user manual or contact Cisco Systems, Inc for 
further information concerning license keys. 

In order to purchase a license, see your local Cisco reseller. If you purchased a license and need a replacement, send an email to licensing@cisco.com.

Q. What details does the show dos command provide?

A. Cisco CSS can display details about the most recent attack events, which include:

  • Source and destination IP addresses

  • The event type

  • Total occurrences

If multiple attacks occur with the same Denial of Service (DoS) type and source and destination address, there is an attempt to merge them as one event. This merge reduces the display of events.

Issue the show dos command in order to display:

  • The total number of attacks since the boot of the CSS

  • The types of attacks and the maximum number of these attacks per second

  • The first and last occurrence of an attack

This example shows the output from the show dos command:

CSS50-1#show dos
Denial of Service Attack Summary: 
Total Attacks: 0 
SYN Attacks:                          0 Maximum per second:                   0 
LAND Attacks:                         0 Maximum per second:                   0 
Zero Port Attacks:                    0 Maximum per second:                   0 
Illegal Src Attacks:                  0 Maximum per second:                   0 
Illegal Dst Attacks:                  0 Maximum per second:                   0 
Smurf Attacks:                        0 Maximum per second:                   0 
  

No attacks detected

This list provides a brief description of each of the fields that the command displays:

  • Total Attacks—The total number of DoS attacks that were detected since the boot of the box. You can find a description of the type of attacks that appear in the list, along with the number of occurrences, below.

  • SYN Attacks—The TCP connections that a source initiates but that are not followed with an acknowledgement frame in order to complete the three-way TCP handshake.

  • LAND Attacks—Any packets that have identical source and destination addresses. The CSS does not allow internal IP addresses to be the source address of a flow. Also, the CSS does not allow the source and destination addresses of frames to be equal.

  • Zero Port Attacks—Frames that contain source or destination TCP or User Datagram Protocol (UDP) ports that are equal to zero.

    Note: Older SmartBits software can send frames that contain source or destination ports equal to zero. The CSS logs them as DoS attacks and drops these frames.

  • Illegal Src Attacks—Illegal source addresses.

  • Illegal Dst Attacks—Illegal destination addresses.

  • Smurf Attacks—Pings with a broadcast destination address. The CSS does not allow directed broadcasts by default. A Smurf Attack uses an Internet Control Message Protocol (ICMP) echo to a broadcast address. The CSS can block access to UDP echo ports via Access Control Lists (ACLs).

  • Maximum per second—The maximum number of events per second. Use the maximum-events-per-second information to set Simple Network Management Protocol (SNMP) trap threshold values.

    Note: The maximum number of events per second is the maximum per Small Form Factor Pluggable (SFP). For a CSS 11800, for example, which can have up to four SFPs, the maximum rate per second can be as high as four times the number that appears in the display.

    Note: Another FAQ asks if you can disable DoS protection on the CSS. The answer is no. The DoS protection is part of the flow admission process. The intention of DoS protection is to protect the resources in the CSS as well as the servers behind the CSS. DoS is not a configurable item. The intention is for DoS to be transparent when protocols work correctly. The flow setup process deeply involves the DoS features. The features help the CSS conserve fast path resources and protect devices that the CSS reaches. The features are always present in software version 3.0 and later.

Also consider the setup of certain SNMP traps for the detection of possible DoS attacks. The available traps are:

  • snmp trap-type enterprise—In order to enable SNMP enterprise traps and configure trap types, issue the snmp trap-type enterprise command. Issue the no snmp trap-type enterprise command in order to disable all traps. You must enable enterprise traps before you configure an enterprise trap option. You can enable the CSS to generate enterprise traps when DoS attack events occur, a login fails, or a CSS service transitions state.

  • dos_attack_type —Generates SNMP enterprise traps when a DoS attack event occurs. One trap generation occurs each second when the number of attacks during that second exceeds the threshold for the DoS attack-type configuration. The options are:

    • dos-illegal-attack—Generates traps for illegal addresses, either source or destination. Illegal addresses are:

      • Loopback source addresses

      • Broadcast source addresses

      • Loopback destination addresses

      • Multicast source addresses

      • Source addresses that you own

      The default trap threshold for this type of attack is one per second.

    • dos-land-attack—Generates traps for packets that have identical source and destination addresses. The default trap threshold for this type of attack is one per second.

    • dos-ping-attack—Generates traps when the number of pings exceeds the threshold value. The default trap threshold for this type of attack is 30 per second.

      Note: This option does not track pings of death DoS attacks.

    • dos-smurf-attack—Generates traps when the number of pings with a broadcast destination address exceeds the threshold value. The default trap threshold for this type of attack is one per second.

    • dos-syn-attack—Generates traps when the number of TCP connections that a source initiates but that are not followed with an acknowledgement frame to complete the three-way TCP handshake exceeds the threshold value. The default trap threshold for this type of attack is 10 per second.

Q. Can I turn off the Denial of Service (DoS) protection feature on the CSS line of switches?

A. In the current line of software for the CSS (Cisco WebNS), there is no option to disable the DoS protection feature.

Q. Can I turn off the Denial of Service (DoS) protection counters?

A. There is no option to disable the counters that log DoS/SYN attacks.

Note: For more information on DoS and SYN attacks, see the response to the FAQ What details does the show dos command provide?.

Q. How do I use port ranges in access lists?

A. The use of port ranges in an Access Control List (ACL) helps simplify the number of ACLs that you configure, given a situation in which you want to block user access for some TCP/User Datagram Protocol (UDP) ports. For example, suppose that you want to block ports 20 through 23 for all users who come into the box from outside your network. First, assume that the outside network or public side of the CSS is in VLAN 2. Also assume that the internal or server side of the network is on VLAN 1. The ACL configuration is:

acl 1 
clause 10 deny any any destination range 20 23  

!--- This clause blocks.
 
clause 20 permit any any destination any   

!--- This clause allows everything else.

apply circuit-(VLAN2) 
acl 
clause 10 permit any any destination any 
apply circuit-(VLAN1)

Related Information

Updated: Apr 19, 2006
Document ID: 25901