Release Notes for SSG-MWAM Release 1.1 with Cisco IOS Release 12.3(5a)B5

SC-SVC-SSD-10=—Service Selection Gateway Layer 2 Tunneling Protocol dial out license

These release notes include important information and caveats for Cisco SSG-MWAM Release 1.1, which provides the Service Selection Gateway (SSG) feature on the Multi-processor WAN Application Module (MWAM) using Cisco IOS Release 12.3(5a)B5.

Introduction

Cisco SSG-MWAM Release 1.1 implements the SSG on the Multi-processor WAN Application Module (MWAM). Cisco SSG-MWAM Release 1.1 increases session density and enhances interoperability with other products based on the Catalyst 6500/Cisco 7600 series platform.

Multi-processor WAN Application Module

The MWAM provides three processor complexes with dual processors used in two of the complexes and a single processor used in the remaining processor complex. This architecture provides five SSGs (see Figure 1) on one module. In addition, each Catalyst 6500/Cisco 7600 chassis can be populated with multiple MWAMs to enable a large number of subscribers to access network services under SSG control.

The MWAM does not provide external ports but is connected to the switch fabric in the Catalyst 6500/Cisco 7600 chassis. An internal Gigabit Ethernet port provides an interface between each processor complex and the Supervisor module. Virtual Local Area Networks (VLANs) direct traffic from external ports via the Supervisor module to each SSG instance.

The software image that provides the SSG feature is downloaded through the Supervisor module and distributed to each processor complex on the MWAM(s). The same image is installed on all the processors in the MWAM.

Service Selection Gateway

The SSG is a Cisco IOS software feature module that enables service providers to create new revenue-generating opportunities by offering on-demand services. The SSG provides Remote Authentication Dial-in User Service (RADIUS) authentication and accounting for user-interactive policy routing to different IP destinations. This improves flexibility and convenience for subscribers, including the ability to log on to multiple services simultaneously, and enables service providers to bill subscribers based on connection time and services used, rather than charging a flat rate.

Traffic from the mobile user is addressed to an SSG on the MWAM. The request for access is forwarded to the Authentication, Authorization, and Accounting (AAA) server, and the user is authenticated and authorized to access the services defined in a user profile. Then data traffic is exchanged between the user and servers in the service network. Each network is defined with its own VLAN, and all SSGs on the MWAM access the same VLANs to receive and send data.

For more information about the features available in the SSG, refer to the following URL:

System Requirements

Hardware

A Hardware-Software Compatibility Matrix is available on CCO for users with CCO login accounts. This matrix allows users to search for supported hardware components by entering a Cisco platform and IOS Release. The Hardware-Software Compatibility Matrix tool is available at the following URL:

Software

Memory

The MWAM provides two complexes that are equipped with 1 GB memory shared between two processors (512 MB each). The remaining processor complex, the one with only one processor, is equipped with 512 MB memory. The total memory capacity for the MWAM is 2.5 GB.

Determining the Software Version

To determine the version of Cisco IOS software running on your MWAM, log in to the router on one of the MWAM processors and enter the show version EXEC command:

Cisco Internetwork Operating System Software

IOS (tm) MWAM Software (MWAM-G4JS-M), Version 12.3(5a)B3, RELEASE SOFTWARE (fc1)

TAC Support: http://www.cisco.com/tac

Copyright (c) 1986-2002 by cisco Systems, Inc.

Upgrading to a New Software Release

For information on upgrading to a new software release, see the product bulletin Cisco IOS Software Upgrade Ordering Instructions located at:

For information about upgrading SSG images on the MWAM, refer to the Multiprocessor WAN Application Module User Guide:

Note The image download process loads the IOS image onto the three processor complexes on the MWAM.

New and Changed Information

The following is a list of the new hardware and software features supported by the MWAM on the Catalyst 6500/Cisco 7600 family for Cisco IOS Release 12.3B.

New Hardware Features in Cisco IOS Release 12.3(5a)B5

There are no new hardware features supported by the Catalyst 6500/Cisco 7600 family for Cisco IOS Release 12.3(5a)B5.

New Software Features in Cisco IOS Release 12.3(5a)B5

There are no new software features supported by the Catalyst 6500/Cisco 7600 family for Cisco IOS Release 12.3(5a)B5.

New Hardware Features in Cisco IOS Release 12.3(5a)B4

There are no new hardware features supported by the Catalyst 6500/Cisco 7600 family for Cisco IOS Release 12.3(5a)B4.

New Software Features in Cisco IOS Release 12.3(5a)B4

There are no new software features supported by the Catalyst 6500/Cisco 7600 family for Cisco IOS Release 12.3(5a)B4.

New Hardware Features in Cisco IOS Release 12.3(5a)B3

There are no new hardware features supported by the Catalyst 6500/Cisco 7600 family for Cisco IOS Release 12.3(5a)B3.

New Software Features in Cisco IOS Release 12.3(5a)B3

There are no new software features supported by the Catalyst 6500/Cisco 7600 family for Cisco IOS Release 12.3(5a)B3.

New Hardware Features in Cisco IOS Release 12.3(5a)B2

There are no new hardware features supported by the Catalyst 6500/Cisco 7600 family for Cisco IOS Release 12.3(5a)B2.

New Software Features in Cisco IOS Release 12.3(5a)B2

There are no new software features supported by the Catalyst 6500/Cisco 7600 family for Cisco IOS Release 12.3(5a)B2.

New Hardware Features in Cisco IOS Release 12.3(5a)B1

There are no new hardware features supported by the Catalyst 6500/Cisco 7600 family for Cisco IOS Release 12.3(5a)B1.

New Software Features in Cisco IOS Release 12.3(5a)B1

There are no new software features supported by the Catalyst 6500/Cisco 7600 family for Cisco IOS Release 12.3(5a)B1.

New Hardware Features in Cisco IOS Release 12.3(5a)B

There are no new hardware features supported by the Catalyst 6500/Cisco 7600 family for Cisco IOS Release 12.3(5a)B.

New Software Features in Cisco IOS Release 12.3(5a)B

The following new software features are supported by the Catalyst 6500/Cisco 7600 family for Cisco IOS Release 12.3(5a)B.

These features are provided by a new release of the application partition on the MWAM. For more information, go to the following URL:

New Hardware Features in Cisco IOS Release 12.3(3)B1

There are no new hardware features supported by the Catalyst 6500/Cisco 7600 family for Cisco IOS Release 12.3(3)B1.

New Software Features in Cisco IOS Release 12.3(3)B1

There are no new software features supported by the Catalyst 6500/Cisco 7600 family for Cisco IOS Release 12.3(3)B1.

New Hardware Features in Cisco IOS Release 12.3(3)B

There are no new hardware features supported by the Catalyst 6500/Cisco 7600 family for Cisco IOS Release 12.3(3)B.

New Software Features in Cisco IOS Release 12.3(3)B

The following new software features are supported by the Catalyst 6500/Cisco 7600 family for Cisco IOS Release 12.3(3)B:

Attribute Screening for Access Requests

Attribute Screening for Access Requests features allows you to configure your network access server (NAS) to filter attributes in outbound Access Requests to the RADIUS server for purposes of authentication or authorization.

RADIUS NAS-IP-Address Attribute Configurability

The RADIUS NAS-IP-Address Attribute Configurability feature allows you to configure an arbitrary IP address to be used as RADIUS attribute 4, NAS-IP-Address, without changing the source IP address in the IP header of the RADIUS packets. This feature may be used for situations in which service providers are using a cluster of small network access servers (NASs) to simulate a large NAS to improve scalability. This feature allows the NASs to behave as a single RADIUS client from the perspective of the RADIUS server.

SSG Default DNS Redirection

The SSG Default DNS Redirection feature allows a default Domain Name System (DNS) domain to be configured in a service profile. When a default DNS domain is configured, all DNS queries that do not match a service with a specific domain name will be redirected to the DNS server for a default service.

SSG Enhancements

SSG Enhancements describes Layer 2 Tunneling Protocol (L2TP) enhancements for authentication, service logon, and the interface between the Service Selection Gateway (SSG) and the Subscriber Edge Services Manager (SESM). For Release 12.3(3)B, SSG enhancements include a new Account-Info vendor specific attribute (VSA), Account-Accept VSA, and Service-Accept VSA.

The SSG interacts with the SESM, through a Remote Authentication Dial-in User Service (RADIUS) interface. SSG Enhancements describe the enhancements to the RADIUS interface to allow a separate Mobile Station ISDN Number (MSISDN) and Challenge Handshake Authentication Protocol (CHAP) for service logon. The SSG Enhancements documentation also describes error codes in the SSG response to the SESM.

SSG Permanent TCP Redirection

The SSG Permanent TCP Redirection feature enables Service Selection Gateway (SSG), in conjunction with Cisco Subscriber Edge Services Manager (SESM), to provide service selection support to users whose web browsers are configured with HTTP proxy servers. This feature supports plug-and-play functionality in Public Wireless LANs.

SSG TCP Redirect Enhancements

The TCP Redirect feature is enhanced to allow access lists to be associated with server groups. This enhancement can be used to limit the kind of traffic that is redirected based on the source or destination IP address and/or TCP ports. It can also be used to redirect different sets of users to different dashboards for unauthenticated user and unauthorized service redirection.

For more information, see the SSG TCP Redirect Enhancements feature at the following URL:

SSG Transparent Auto-Logon

The SSG Transparent Auto-Logon (TAL) feature enables the Service Selection Gateway (SSG) to authenticate/authorize users based on IP packets received from the user. SSG authorizes users by using information from the Authentication, Authorization, and Accounting (AAA) server when a first IP packet is received from the user.

Users can be activated on SSG through Web-based login procedures using Service Edge Subscriber Management (SESM), RADIUS Proxy, and PPP session termination. The Transparent Auto-Logon feature provides an additional activation method. Transparent Auto-Logon provides SSG services to a user who is authorized based on the source IP address of packets received on a downlink interface of SSG, without any previous authentication phase.

For more information on the Transparent Auto-Logon feature, see the following URL:

New Hardware Features in Cisco IOS Release 12.3(1a)BW

The following new hardware feature is supported by the Catalyst 6500/Cisco 7600 family for Cisco IOS Release 12.3(1a)BW:

MWAM on Catalyst 6500/Cisco 7600 Platform

The MWAM is built on a base card-to-daughter card configuration (see Figure 1). It provides three SiByte (700MHz) processor complexes. Two of the processor complexes enable dual processors while the third processor complex enables only one processor because of the memory configuration.

Each SiByte complex has a 1 Gigabit Ethernet (GE) interface to the switch fabric. This connection appears as a GE interface from the Supervisor module.

The MWAM connects to the Catalyst 6500/Cisco 7600 bus for data and control traffic.

New Software Features in Cisco IOS Release 12.3(1a)BW

The following new software features are supported by the Catalyst 6500/Cisco 7600 family for Cisco IOS Release 12.3(1a)BW.

EAP SIM Enhancements

•

AZR issue: SSG to cleanup the active hosts (EAP-SIM and SESM) users on receiving an Accounting On/Off from AZR due to a reboot. This is needed to close a security hole where an illegal user can hijack the session of a valid user by using the IP address of the valid user after the AZR reboot.

•

SESM reconnect for EAP-SIM users: This requires that EAP-SIM users access the SESM and perform an Account Logoff. Subsequent to the logoff they can access the SESM and do an account logon again.

IP Pool Backup

The IP Pool Backup feature introduces two new interface configuration commands, peer pool backup and peer pool static, which allow you to define alternate sources for IP address pools in the event the original address pool is not present or is exhausted.

The peer pool backup command is useful in large-scale dial-out environments with large numbers of independently controlled authentication, authorization, and accounting (AAA) servers that can make it difficult for the network access server (NAS) to provide proper IP address pool resolution in the following cases:

•

A new pool name is introduced by one of the AAA servers before that pool is set up on the NAS.

•

An existing local pool becomes exhausted, but the owner of that AAA server has other pools that would be acceptable as an IP address source.

The peer pool backup command uses the local pool names configured with the peer default ip address pool interface configuration command to supplement the pool names supplied by AAA. The problems of pool name resolution and specific local pool exhaustion can be solved by configuring backup pool names on a per-interface basis using the peer default ip address pool and peer pool backup interface configuration commands.

The peer pool static command controls attempts by the pool software to load dynamic pools in response to a pool request from a specific interface. These dynamic pools are loaded at system startup and refreshed whenever a pool name not configured on the NAS is specified for IP address allocation. Because the behavior of the NAS in response to a missing pool name can be changed using the peer pool backup interface configuration command, you can use the peer pool static command to control attempts to load all dynamic pools when the AAA-supplied pool name is not an existing local pool name.

Multilink PPP Minimum Links Mandatory

Multilink PPP allows multiple PPP links to be established in parallel to the same destination. Multilink PPP is often used with dialup lines or ISDN connections to easily increase the amount of bandwidth between points.

With the introduction of the Multilink PPP Minimum Links Mandatory feature, you can configure the minimum number of links in a Multilink PPP (MLP) bundle required to keep that bundle active by entering the ppp multilink min-links links mandatory command. When you configure this command, all Network Control Protocols (NCPs) for an MLP bundle are disabled until the MLP bundle has the required minimum number of links. When a new link is added to the MLP bundle that brings the number of links up to the required minimum number of links, the NCPs are activated for the MLP bundle. When a link is removed from an MLP bundle, and the number of links falls below the required minimum number of links for that MLP bundle, the NCPs are disabled for that MLP bundle.

Using the PPPoE Session Limit Per NAS Port feature, you can limit the number of sessions on a specific virtual circuit (VC) or VLAN configured on an L2TP access concentrator (LAC). The NAS port is either an ATM VC or a configured VLAN ID.

The PPPoE session limit per NAS port is maintained in a RADIUS server customer profile database. This customer profile database is connected to a LAC and is separate from the RADIUS server that the LAC and L2TP Network Server (LNS) use for the authentication and authorization of incoming users. When the customer profile database receives a pre-authorization request from the LAC, it sends the PPPoE per NAS port session limit to the LAC.

The LAC sends a pre-authorization request to the customer profile database when the LAC is configured for Subscriber Service Switch (SSS) pre-authorization. Configure the LAC for SSS pre-authorization using the sss-subscriber access pppoe pre-authorize command. When the LAC receives the PPPoE per NAS port session limit from the customer profile database, the LAC compares the PPPoE per NAS port session limit to the number of sessions currently on the NAS port. The LAC then decides whether to accept or reject the current call based upon the configured PPoE per NAS port session limit and the number of calls currently on the NAS port.

You can configure other types of session limits on the LAC including session limit per VC, per VLAN, per MAC, or a global session limit for the LAC. When PPPoE Session Limit Per NAS Port is enabled (that is, when you have enabled SSS pre-authorization on the LAC), local configurations for session limit per VC and per VLAN are overwritten by the PPPoE per NAS port session limit downloaded from the customer profile database. Configured session limits per VC and per VLAN serve as backups in case of a PPPoE per NAS port session limit download failure.

The customer profile database consists of user profiles for each user connected to the LAC. Each user profile contains the NAS-IP-Address (Attribute #4) and the NAS-Port-ID (Attribute #5.) When the LAC is configured for SSS pre-authorization, it queries the customer profile database using the username. When a match is found in the customer profile database, the customer profile database sends the PPPoE per NAS port session limit in the user profile. The PPPoE per NAS port session limit is defined in the username as a Cisco AVpair.

RFC-2867 RADIUS Tunnel Accounting

The RFC-2867 RADIUS Tunnel Accounting feature introduces six new RADIUS accounting types that are used with the RADIUS accounting attribute Acct-Status-Type (attribute 40), which indicates whether an accounting request marks the beginning of user service (start) or the end (stop). These new accounting types are designed to support the provision of compulsory tunneling in dialup networks; that is, these attribute types allow you to better track tunnel status changes.

This feature also introduces two new commands—vpdn session accounting network (tunnel-link-type records) and vpdn tunnel accounting network (tunnel-type records)—that help identify the following events:

Note The first two events are tunnel-type accounting records: authentication, authorization, and accounting (AAA) sends Tunnel-Start, Tunnel-Stop, or Tunnel-Reject accounting records to the RADIUS server. The next two events are tunnel-link-type accounting records: AAA sends Tunnel-Link-Start, Tunnel-Link-Stop, or Tunnel-Link-Reject accounting records to the RADIUS server.

Note The accounting types are divided into two separate tunnel types so users can decide if they want tunnel type, tunnel-link type, or both types of accounting.

Service Selection Gateway

Service Selection Gateway (SSG) is a switching solution for service providers who offer intranet, extranet, and Internet connections to subscribers using broadband access technology such as xDSL, cable modems, or wireless to allow simultaneous access to network services.

SSG Autologoff Enhancement

The SSG Autologoff Enhancement feature configures Service Selection Gateway (SSG) to check the MAC address of a host each time that SSG performs an Address Resolution Protocol (ARP) ping. If SSG finds that the MAC address of the host has changed, SSG automatically initiates the logoff of that host. This prevents unauthorized reuse of IP addresses (spoofing). SSG MAC address checking also detects the assignment of a host IP address to a different host before the original hosts initiates a logoff and clears its host object. This prevents session reuse by a second host.

The ARP is an Internet protocol used to map IP addresses to MAC addresses in directly connected devices. A router that uses ARP will broadcast ARP requests for IP address information. When an IP address is successfully associated with a MAC address, the router stores the information in the ARP cache.

When SSG Autologoff is configured to use ARP ping, SSG periodically checks the ARP cache tables. If a table entry for a host is found, SSG forces ARP to refresh the entry and checks the entry again after a configured interval. If a table entry is not found, SSG initiates autologoff for the host. However, if any data traffic to or from the host occurred during the interval, SSG does not ping the host because the reachability of the host during that interval was established by the data traffic.

When SSG MAC address checking is configured, SSG checks the MAC address of a host when an ARP ping is performed. If SSG detects a different host MAC address, it initiates an automatic logoff of that host.

Note ARP ping should be used only in deployment scenarios in which all hosts are directly connected to SSG through a broadcast interface such as an Ethernet interface or a bridged interface such as a routed bridge encapsulation (RBE) or integrated routing and bridging (IRB) interface.

ARP request packets are smaller than Internet Control Message Protocol (ICMP) ping packets, so it is recommended that you configure SSG Autologoff to use ARP ping in scenarios where hosts are directly connected.

SSG Complete ID

SSG Complete ID provides enhancements to the current interaction mechanism that is used between SSG and SESM, allowing SSG to pass along the following additional information:

This allows SESM to offer greater customization of Web portals, specifically by locations. Each hotspot can now have its own branded portal.

SSG EAP Transparency

The SSG EAP Transparency feature allows SSG to transparently pass EAP-SIM, EAP-TLS and Cisco LEAP authentication.

SSG Open Garden Configuration Enhancements

The Service Selection Gateway (SSG) is an IOS feature and implements layer 3 service selection through selective routing of IP packets to destination networks on a per subscriber basis. Out of the many features SSG has, Open Garden is one of the features, which is very useful for service providers to provide trial-based services to the customers.

An open garden is a collection of web sites that a user can access as long as the user has physical access to the network. The user doesn't need to provide any authentication information before accessing the Web sites in the open garden.

Currently, SSG open garden services can be configured/managed on the router itself, even though they are similar to normal SSG (subscribed) services. The modifications being proposed will allow open garden services to be defined and managed on the RADIUS server as well.

SSG L2TP Dialout

The SSG L2TP Dialout feature enhances SSG tunnel services and provides a dialout facility to users. Many Small Office Home Offices (SOHOs) use the Public Switched Telephone Network (PSTN) to access their intranet. SSG L2TP provides mobile users with a way to securely connect to their SOHO through the PSTN.

To provide SSG L2TP Dialout, SSG requires a digital number identification service (DNIS) number for the SOHO to which the user wants to connect, the address of the L2TP Access Concentrator (LAC) closest to the SOHO, and configured tunnel parameters to establish a tunnel to the LAC.

Users can access SSG L2TP Dialout by selecting the dialout service using Cisco Subscriber Edge Services Manager (SESM) from the list of subscribed services or by using a structured username. The user must provide the DNIS number when using either method of connecting to the dialout service.

SSG Prepaid Enhancements

The SSG Prepaid feature allows SSG to check a subscriber's available credit to determine whether to connect the subscriber to a service and how long the connection can last. The subscriber's credit is administered by the billing server as a series of quotas representing either a duration of use (in seconds) or an allowable data volume (in bytes). A quota is an allotment of available credit.

To obtain the first quota for a connection, SSG submits an authorization request to the authentication, authorization, and accounting (AAA) server. The AAA server contacts the prepaid billing server, which forwards the quota values to SSG. SSG then monitors the connection to track the quota usage. When the quota runs out, SSG performs reauthorization. During reauthorization, the billing server may provide SSG with an additional quota if there is available credit. If no further quota is provided, SSG logs off the user.

SSG Prepaid Enhancements introduces prepaid tariff switching, simultaneous volume and time based prepaid billing, and postpaid tariff Switching.

SSG Prepaid Idle Timeout

The SSG Prepaid Idle Timeout feature enhances the SSG Prepaid feature by enabling SSG to return residual quota to the billing server from services that a user is logged into but not actively using. The quota that is returned to the billing center can be applied to the quota for the services the user is actively using.

When SSG is configured for SSG Prepaid Idle Timeout, a user's connection to services can be open even when the billing server returns a zero quota, but the connection's status is dependent on the combination of the quota and the idle timeout value returned. Depending on the connection service, SSG requests the quota for a connection from the billing server once the user starts using a particular service, when the user runs out of quota, or after the configured idle timeout value has expired.

The SSG Prepaid Idle Timeout feature enhances handling of a returned zero quota from the billing server. If a billing server returns a zero quota, and non-zero idle timeout, this indicates that a user has run out of credit for a service. When a user runs out of credit for a service, the user is redirected to the billing server to replenish the quota. When the user is redirected to the billing server, the user's connection to the original service or services is retained. Although the connection remains up, any traffic passing through the connection is dropped. This enables a user to replenish quota on the billing server without losing connections to services or having to perform additional service logons.

Using the SSG Prepaid Idle Timeout feature, you can configure SSG to reauthorize a user before the user completely consumes the allocated quota. You can also configure SSG to not pass traffic during reauthorization. This prevents revenue leaks in the event that the billing server returns a zero quota for the user. Without the SSG Prepaid Idle Timeout feature, traffic passed during reauthorization represents a revenue leak if the billing server returns a zero quota for the user. You can prevent this type of revenue leak by configuring a threshold value, causing SSG to reauthorize a user's connection before the user completely consumes the allocated quota for a service.

SSG Prepaid Idle Timeout enhances SSG to inform the billing server upon any connection failure. This enables the billing server to free quota that was reserved for the connection that failed and to apply this quota immediately to some other active connection.

SSG Proxy for CDMA2000

The SSG Proxy for CDMA2000 extends the functionality of the existing SSG RADIUS Proxy so that it may be used in CDMA2000 networks.

When used in a CDMA2000 network, SSG provides RADIUS proxy services to the Packet Data Serving Node (PDSN) and the Home Agent (HA) for both Simple IP and Mobile IP authentication. SSG also provides service selection management and policy-based traffic direction for subscribers.

SSG Proxy for CDMA2000, used with Cisco Subscriber Edge Services Manager (SESM), provides users with on-demand services and service providers with service management and subscriber management.

SSG Proxy for CDMA2000 supports time- and volume-based usage accounting for Simple IP and Mobile IP sessions. Prepaid and postpaid services are supported. Host accounting records can be sent to multiple network elements including Content Service Gateways (CSGs), Content Optimization Engines (COEs), and Wireless Application Protocol (WAP) gateways.

Code Division Multiple Access (CDMA) is a digital spread-spectrum modulation technique used mainly with personal communications devices such as mobile phones. CDMA digitizes the conversation and tags it with a special frequency code. The data is then scattered across the frequency band in a pseudorandom pattern. The receiving device is instructed to decipher only the data corresponding to a particular code to reconstruct the signal.

CDMA2000 Radius Transmission Technology (RTT) is a wideband, spread-spectrum radio interface that uses CDMA technology to satisfy the needs of Third generation (3G) wireless communication systems. CDMA2000 is backward compatible with CDMA.

For more information about CDMA2000, refer to the "CDMA2000 Overview" knowledge byte on the Mobile Wireless Knowledge Bytes web page.

When used in a CDMA2000 environment, SSG acts as a RADIUS proxy to the Packet Data Serving Node (PDSN) and to the Home Agent for Simple IP authentication. SSG sets up a host object for the following three different access modes:

•

PAP/CHAP authentication. In this mode, Password Authentication Protocol/ Challenge Handshake Authentication Protocol (PAP/CHAP) is performed during PPP setup and the NAI is received from a mobile node (MN).

•

MSID-Based Access. In this mode, the MN does not negotiate CHAP or PAP and no Network Access Identifier (NAI) is received by the PDSN. The PDSN does not perform additional authentication. PDSN constructs an NAI based on the MSID and generates accounting records. Because a user password is not available from the MN, a globally configured password is used as the service password.

•

MSID-Based Access-Cisco Variant. In this mode, a Cisco PDSN supports MSID-based access by using a realm retrieved from the RADIUS server. This realm is retrieved during an extra authentication phase with the RADIUS server.

SSG operating in a CDMA2000 network correlates Accounting-Start and Accounting-Stop requests. A PDSN may send out many Accounting-Start and Accounting-Stop requests during a session. These Accounting-Start and Accounting-Stop requests can be generated by PDSN hand-off, Packet Control Function (PCF) hand-off, interim accounting, and time-of-date accounting. SSG terminates a session only when it receives an Accounting-Stop request with the 3GPP2-Session-Continue VSA set to "FALSE" or if a subsequent Accounting-Start request is not received within a configured timeout. PPP renegotiation during a PDSN hand-off is treated as a new session.

In SSG Proxy for CDMA2000 for Simple IP, the end-user IP address may be assigned statically by the PDSN, RADIUS server, or SSG. The end-user IP address can also be assigned directly from the autodomain service.

Network Address Translation (NAT) is automatically performed when necessary. NAT is generally necessary when IP address assignment is performed by any mechanism other than directly from the autodomain service (which may be a VPN). You can also configure SSG to always use NAT.

If the user profile contains Cisco Attribute-Value (AV)-pairs of Virtual Private Dialup Network (VPDN) attributes, SSG initiates Layer 2 Tunneling Protocol (L2TP) VPN.

For Mobile IP, SSG functions as the RADIUS proxy for both PDSN and the HA. SSG proxies PPP PAP or CHAP and Mobile Node (MN)/Foreign Agent (FA) CHAP authentication. SSG Proxy for CDMA2000 for Mobile IP can assign IP addresses statically by the PDSN, RADIUS server, or SSG. The end user IP address can also be assigned directly from the autodomain service.

Home Agent-Mobile Node (HA-MN) authentication and reverse tunneling must be enabled so that SSG can create host objects for Mobile IP sessions based on proxied RADIUS packets received from the HA.

The Home Agent must generate RADIUS accounting packets so that SSG can discover the user IP address and detect the termination of the session. Multiple Mobile IP sessions with the same NAI are supported. RADIUS packets must contain the Accounting-Session-ID attribute to be associated with the correct user session. SSG correlates RADIUS packets from the PDSN in order to obtain MSID information for a host object of a Mobile IP session.

SSG can set up a host object either with or without PAP/CHAP performed during the original PPP session.

SSG initiates L2TP VPN according to the SSG tunnel service VSAs in the user's profile. If the user profile contains Cisco AV-pairs of VPDN, SSG sets up the L2TP tunnel per these VPDN attributes. SSG removes these AV-pairs when sending the Access-Accept packet back to the PDSN.

•

The RADIUS server selects the local HA or any HA that is configured for session requests. For foreign-user call requests, the AAA server assigns the HA.

•

SSG modifies the fixed HA address received from the RADIUS server to a local HA address. This method can be implemented without making any changes to the RADIUS server configuration. SSG does not modify the HA address for a foreign user. The foreign-user call request is registered with the HA address assigned by the AAA server.

•

The PDSN implements dynamic HA assignment based on detection of the PDSN hand-off.

SSG Proxy for CDMA2000 provides geographical redundancy by copying host object accounting packets and sending them to multiple RADIUS servers.

SSG PTA-MD Exclusion Lists

Beginning in Cisco IOS Release 12.2(8)B, the option of passing the entire structured username in the form `user@service' to PPP for authenticating an SSG request became available. The entire structured username can be passed to PPP through the use of a PTA-MD exclusion list; if an entire structured username should be passed to PPP, the domain (the `@service' portion of the structured username) should be added to a PTA-MD exclusion list. The PTA-MD exclusion list can be configured on the AAA server directly or via the router CLI. Structured usernames are parsed for authentication unless a PTA-MD exclusion list is configured for the particular domain requesting a service.

SSG Range Command for Bind Statements

SSG Range Command for Bind Statements creates a A "range" command for SSG BIND statements. This is useful when provisioning RBE subscribers en masse, as it allows for streamlined provisioning and configuration with a decreased CPU load.

SSG Service Profile Caching

The SSG Service Profile Caching feature enhances the authentication process for SSG services by allowing users to authenticate a service using the service profile cached in SSG.

When SSG Service Profile Caching is not enabled, an authentication, authorization, and accounting (AAA) transaction is required to download a service profile each time an SSG subscriber logs onto the service. The other SSG subscribers already logged onto the service also have their service parameters automatically refreshed as a result of this AAA transaction. In many cases, this automatic refresh causes unnecessary traffic in SSG and on the AAA server.

The SSG Service Profile Caching feature creates a cache of service profiles in SSG. A service profile is downloaded from the AAA server and then stored in the SSG service profile cache as a service-info object. Subsequent SSG subscribers hoping to use that service are authorized by the SSG service profile cache provided that service profile remains in the cache. To ensure that the service profiles in the SSG service profile cache remain updated, the SSG service profile cache automatically refreshes the service profiles by downloading the service profiles from the AAA server at user-configured intervals (the default is every 120 minutes). SSG service profile caches can also be refreshed manually at any time. Service profiles that are not being used by any SSG subscriber are removed from the SSG service profile cache.

SSG Support of NAS Port ID

This feature will carry the NAS-Port attribute in the authentication packet. This will allow the authentication server to use consistent policies while authenticating PPPoX users and RFC1483 users. Currently, NAS-Port attribute is sent only for PPPoX users.

With this feature, SSG will send nas-port information for certain IP users in the authentication-request and accounting-request packets.

SSG Suppression of Unused Accounting Records

The SSG Suppression of Unused Accounting Records feature provides the ability to turn off those accounting records that are not needed on the router.

SSG Unconfig

The SSG Unconfig feature enhances your ability to disable SSG at any time and releases the data structures and system resources created by SSG when SSG is unconfigured.

The SSG Unconfig feature enhances several IOS commands to delete all host objects, delete a range of host objects. You can also delete all service objects or connection objects. The show ssg host command has been enhanced to display information about an interface and its IP address when Host-Key mode is enabled on that interface.

When you enable SSG, the SSG subsystem in IOS acquires system resources that are never released, even after you disable SSG. The SSG Unconfig feature enables you to release and clean up system resources when SSG is not in use by entering the no ssg enable force-cleanup command.

SSG Unique Session ID

SSG does not currently support a totally unique accounting session ID in the RADIUS accounting records. The SSG Unique Session ID feature provides a unique format in the RADIUS accounting records in order to be compatible with a customer's existing backend billing systems.

Performance

Each SSG instance on the MWAM is an individual router. Because the MWAM supports five SSGs, it provides five times the session density (that is, number of user sessions) of the NPE 400 7200/7400 platform. In addition, the MWAM processors provide twice the throughput of processors used in the NPE 400 7200/7400 platform. Overall, the MWAM improves SSG throughput by 5-10 times that of the NPE 400 7200/7400 platform.

External Interfaces

External physical interfaces provided by the supported platforms are not visible to the SSG software. This is an important advantage of the MWAM implementation when compared to the Cisco 7200/7400 platform. The MWAM implementation protects the SSG from interface and link failures. As long as the platform provides redundant links to other system components (for example, GGSN, AAA servers), the SSG configuration is not affected and its operation is maintained.

IP Address Management

The IP address management for the SSG on the MWAM is the same as the Cisco 7200/7400 platform with one exception: virtual subinterfaces (VLANs) are required for uplink, downlink, and network management paths.

Each SSG on the MWAM is configured with its own IP addresses including addresses for user traffic, RADIUS client function, and network management.

Reliability/Availability

This section provides analysis of reliability/availability of the SSG on the MWAM in the Catalyst 6500/Cisco 7600 chassis in context with other Cisco features. The following features are considered:

•

FWLB on Supervisor module or CSM—Provides the return traffic path through the same SSG that forwarded the service request

The Server Load Balancing (SLB) function can be implemented in the Supervisor module to provide RADIUS Load Balancing (RLB) across the SSGs on one or multiple MWAMs. The Content Switching Module (CSM) can be used to provide Firewall Load Balancing (FWLB).

Note While the RLB and FWLB features are not part of the SSG-MWAM Release 1.1, they are described here to demonstrate their use in mobile wireless solutions that include MWAM-based SSGs.

RADIUS Load Balancer

The RLB feature is implemented in the Supervisor module. The RLB feature provides one virtual IP address for all users accessing services and keeps the list of real IP addresses of all SSGs. The RLB feature distributes the upstream traffic between SSGs by using the load-balancing mechanism. It keeps the information about SSG assignment for each user session. When the RLB detects an SSG failure, it directs traffic to another available SSG.

Firewall Load Balancer

The FWLB feature ensures that the downstream traffic to the user is sent to the same SSG that handled the upstream traffic. The FWLB feature tracks all upstream traffic from an SSG to a network server and links the SSG address with the user session. This information is used when the downstream traffic from the server is received. The FWLB feature determines which SSG is handling the user traffic.

The FWLB feature can be implemented in the CSM in the same chassis or in the Supervisor module in different chassis.

System Modules

Each system module in the configuration provides its own degree of reliability/availability.

Two redundant Supervisor modules can be equipped in the same chassis using the Route Processor Redundancy Plus (RPR+) protocol and the RLB. The RLB provides stateful failover in this configuration (that is, user sessions are preserved).

If equipping redundant Supervisor modules in two chassis, the RLB can be configured with Hot Standby Router Protocol (HSRP) between the two RLBs to provide stateful failover (that is, user sessions are maintained).

When configuring the Supervisor module for the FWLB feature, it must be equipped on a different chassis than the one providing the RLB feature. If two chassis are used, the FWLB feature can be configured with HSRP and provide stateful failover.

One or more MWAMs can be equipped using stateless failover (provided by the RLB feature) between SSGs. In a stateless failover, when user sessions are lost, the user must re-authenticate, but service access is not denied.

Two redundant CSMs can be equipped. The FWLB feature is configured with HSRP to provide stateful failover. No user sessions or data packets are lost.

Other service modules can be installed in the same Catalyst 6500/Cisco 7600 chassis that contains the MWAM. For example, to provide advanced content billing, install the Content Services Gateway (CSG).

Configuration Options

The SSG-MWAM Release 1.1 can be implemented in a redundant configuration using one or two chassis with the RLB feature providing the failover mechanism. The MWAM supports the Supervisor module RPR+ feature. This feature enables the MWAM to continue to operate after the active Supervisor fails and the secondary Supervisor takes over.

•

RLB feature on the Supervisor module to distribute load and provide failover for SSGs

•

Failed SSG or MWAM—User sessions are lost, but the traffic is redirected to active SSGs and users can reactivate their sessions

•

Failed Supervisor module—User sessions are preserved because the active RLB synchronizes its state with its backup (stateful failover using RPR+)

For deployments requiring high reliability/availability, multiple MWAMs in two chassis can be used. The two-chassis configuration uses the following components:

•

If one SSG fails, the RLB feature provides failover to another SSG; all sessions on the failed SSG are lost and users must log in again.

–

Supervisor switchover to the standby Supervisor (using RPR+) in the same chassis

•

Failure of active FWLB causes stateful failover to the standby FWLB, maintaining user sessions

MWAM Installation and Configuration

For information on installing the MWAM, configuring it through the Command Line Interface (CLI), and loading or upgrading IOS images on the MWAM, refer to the Multiprocessor WAN Application Module User Guide:

Limitations, Restrictions, and Important Notes

When working with the MWAM, observe the following limitations, restrictions, and important notes:

•

Only five instances of the Cisco IOS image 12.3(5a)B can be loaded onto the MWAM.

•

Session console is provided by TCP connection from the Supervisor module (no direct console).

•

If one processor in a processor complex fails, the second processor also fails, and both processors must be reset.

Caveats

Caveats describe unexpected behavior in Cisco IOS software releases. Severity 1 caveats are the most serious caveats; severity 2 caveats are less serious.

Note If you have an account with CCO, you can use Bug Navigator II to find caveats of any severity for any release. You can reach Bug Navigator II on CCO at Software Center: Cisco IOS Software: Cisco Bug Toolkit: Cisco Bugtool Navigator II, or at http://www.cisco.com/support/bugtools.

For a list of caveats for 12.2(14)ZA2 (and higher), see the release notes at the following URL:

Resolved Caveats—Cisco IOS Release 12.3(5a)B5

All the caveats listed in this section are resolved in Cisco IOS Release 12.3(5a)5. This section describes only severity 1 and 2 caveats and select severity 3 caveats.

Resolved Miscellaneous Caveats

A memory leak may occur in the dead process on a Cisco router, and the memory allocation failures (MALLOCAIL) may be reported in the processor pool. The authentication, authorization, and accounting (AAA) User Identifier (UID) database may leak about 200 MB for each failed EXEC call or VTY session of the internal errors during the initiation process.

This is observed when the EXEC Accounting and Network Accounting are enabled, and when a failure occurs during an EXEC call or a VTY session. The reasons for the EXEC call failure or VTY session failure could be low processor memory on the Cisco router, an internal message processing error, or a timeout during the prompting for a username and password.

Workaround: If this is an option, disable EXEC Accounting and Network Accounting.

Resolved SSG Caveats

SSG uses a duplicate Acct-Session-Id (attribute 44) in a RADIUS accounting packet.

SSG does not update tariff switch information to the users when the user logs-in exactly at tariff switching time.

Resolved Caveats—Cisco IOS Release 12.3(5a)B4

All the caveats listed in this section are resolved in Cisco IOS Release 12.3(5a)4. This section describes only severity 1 and 2 caveats and select severity 3 caveats.

Resolved Miscellaneous Caveats

Cisco routers running Cisco IOS that supports Multiprotocol Label Switching (MPLS) are vulnerable to a Denial of Service (DoS) attack on MPLS disabled interfaces.

The vulnerability is only present in Cisco IOS release trains based on 12.1T,12.2, 12.2T, 12.3 and 12.3T. Releases based on 12.1 mainline, 12.1E and all releases prior to 12.1 are not vulnerable.

This bug is a complementary fix to CSCeb56909 which addresses this vulnerability.

Consequently, the use of the undebug all command in the privileged EXEC mode on any Cisco 3700 routers used as an IPSec Gateway, will stop all traffic going through an encrypted GRE (generic routing encapsulation) or tunnel using CEF (Cisco Express Forwarding) switching.

Workaround: Re-initialize CEF switching by using the no ip cef and then ip cef commands in the global configuration mode or turn off the individual debugs that have been turned on instead of using the undebug all command.

ESP (Encapsulated Security Payload) and AH IPSec (authentication header IP security) connections may be vulnerable to spoofed ICMP (internet connect message protocol) type 3 code 4 packets (that is, packets that are too large but have the DF (don't fragment) bit set). A spoofed ICMP type 3 code 4 packet may cause IPSec to use very low path MTU (maximum transmission unit) values for a flow for the duration of the security association (SA) lifetime.

This is observed when ESP and AH IPSec connections are configured for PMTU (path MTU) discovery.

Workaround: Disable the PMTU discovery by entering the crypto ipsec df-bit clear command in either the global configuration or interface configuration mode.

Description: Reception of certain IPv6 fragments with carefully crafted illegal contents may cause a router running Cisco IOS to reload if it has IPv6 configured. This applies to all versions of Cisco IOS that include support for IPv6.

The system may be protected by installing appropriate access lists to filter all IPv6 fragments destined for the system. For example:

    ipv6 traffic-filter nofragments in

    deny ipv6 any <my address1> undetermined-transport

    deny ipv6 any <my address2> fragments

This must be applied across all interfaces, and must be applied to all IPv6 addresses which the system recognizes as its own.

This will effectively disable reassembly of all IPv6 fragments. Some networks may rely on IPv6 fragmentation, so careful consideration should be given before applying this workaround.We would recommend for customers to upgrade to the fixed IOS release. All IOS releases listed in IPv6 Routing Header Vulnerability Advisory at

Resolved SSG Caveats

This situation is observed on a Cisco platform that runs SSG under a moderate traffic load when the timer that is attached to the timer wheel has a tick value that is a multiple value of the size of the wheel.

Resolved Caveats—Cisco IOS Release 12.3(5a)B3

All the caveats listed in this section are resolved in Cisco IOS Release 12.3(5a)B3. This section describes only severity 1 and 2 caveats and select severity 3 caveats.

Resolved SSG Caveats

A Cisco 7204VXR series router functioning as an L2TP Network Server (LNS) may pause indefinitely because of a bus error when a user disconnects and then reconnects.

This issue is observed when a Cisco 7204VXR series router that is configured with a Network Processing Engine G1 (NPE-G1) under the following conditions:

–

The router functions as an LNS that terminates Layer 2 Tunneling Protocol (L2TP).

–

The output route filters are applied via RADIUS server attributes to the Routing Information Protocol (RIP) routing process.

A Cisco AS5x00 access server may ignore a service-login attribute and start a PPP session. The Cisco AS5x00 may also start a PPP session when the RADIUS Access-Accept reply contains unknown (unsupported) Framed-Protocol attributes.

This issue is observed when a client uses a Password Authentication Protocol (PAP) for authentication.

This issue occurs when both the force-local-chap and terminate-from hostname commands are configured in the same vpdn-group. Only Cisco IOS software version 12.3T is affected.

Workaround: Use the default L2TP VPDN group by deleting the terminate-from hostname command or use Cisco IOS software version 12.3 mainline train.

The Parallel Express Forwarding (PXF) chunk memory is not freed after the routes are removed. The %PXF-2-TALLOCFAIL message may occur and cause more PXF memory leaking.

This issue occurs on a Cisco 7200 series router with an NSE-1 processor board or a Cisco 7401 series router. When the PXF is enabled, adding/removing routes may not free PXF memory.

A RADIUS NAS-Port attribute is not sent in accounting records and access-requests when:

–

The vpdn aaa attribute nas-port vpdn-nas global configuration command is configured

When the IP User sends an Access Zone Router (AZR) Accounting Off command to the Active Host, the SSG does not clear the Host.

This issue occurs when the SSG is acting as a RADIUS server proxy with Extensible Authentication Protocol-Subscriber Identity Module (EAP-SIM) setup.

Resolved MWAM Caveats

When a 7600 chassis with four MWAMs using the centralized configuration storage feature is reloaded, the MWAM reboot task to load the centralized configuration from Supervisor bootflash takes seven minutes.

If an MWAM is configured to perform IP multicast routing, it will not forward unicast traffic.

Resolved Caveats—Cisco IOS Release 12.3(5a)B2

All the caveats listed in this section are resolved in Cisco IOS Release 12.3(5a)B2. This section describes only severity 1 and 2 caveats and select severity 3 caveats.

Resolved SSG Caveats

–

A Cisco IOS router is configured for VPDN LNS and PPP authentication and authorization via a RADIUS server.

–

The RADIUS server user profile contains the Framed-Route attribute to install a per-user IP Route on the LNS.

Workaround: Clear the Virtual-Access interface (on which the PPP user is terminated) after IPCP renegotiation with the clear interface virtual-access x command.

When TCP connections are configured for PMTU discovery (disabled on router by default), the TCP connections may become vulnerable to spoofed ICMP packets. The spoofed ICMP packet may cause the TCP connection to use a very low segment size for 10 minutes at a time.

Downstream packets from open garden service may not be properly process switched. DNS packets are process switched in SSG, so the DNS replies may not reach the client.

This issue occurs when an internet service is bound to the same interface as the open garden service and an unauthenticated user accesses open garden service.

When multiple users simultaneously log on to and later log off from an SSG L2TP tunnel service, the Cisco platform may run out of IDBs, thus preventing users from connecting to a new SSG L2TP tunnel service.

This issue is observed when the number of virtual-access interfaces that are in use increases (as seen in the output of the show vtemplate command in the privileged EXEC mode). That is, the old virtual-access interfaces that have not been cleared show a long idle time (as seen in the output of the show user command in the user EXEC mode).

Workaround: Clear the unused virtual-access interfaces with the clear interface virtual-access command in the privileged EXEC mode.

A specifically crafted Transmission Control Protocol (TCP) connection to a Telnet or reverse Telnet port of a Cisco product running Cisco IOS� software may block further Telnet, reverse Telnet, Remote Shell (RSH), Secure Shell (SSH), and in some cases Hypertext Transport Protocol (HTTP) access to the Cisco product. The Telnet, reverse Telnet, RSH and SHH sessions established before exploitation are not affected.

All other Cisco product services will operate normally. Services (such as packet forwarding, routing protocols and all other communication to and through the Cisco product) are not affected.

In the current implementation, the prepaid server splits the subscriber credit into an equal amount of volume for both PRE and POST in QX at the tariff switch, such that the following two quota combinations have not been used up:

During operation, this behavior will show that the "equal splitting of the volume for TS" leads to a load problem, if

–

a subscriber doesn't have enough credit at the time of the tariff switch and the subscriber just needs a quota (a lot of re-authorization takes place)

In the upcoming deployment, there will be a tariff every hour for every service with the introduction of HBR and DBR (in the current implementation, the TS occurs once a day at midnight). To avoid the "re-auth vibration" plans are to apply the two quota combinations as shown in the upcoming deployment. If a subscriber has a low credit time before the TS, the prepaid server will grant all of its credit to PRE and QX with Post=0. The SG sends a re-authorization request to the prepaid server when the tariff switch happens (if the PRE has not been consumed before TS).

But now we have a problem, SSG doesn't send the QB attribute in the re-authorization packets for the two cases above. Without QB (time stamp of TS), the prepaid server doesn't know the time point of the tariff switch, as if the TS would not have taken place.

When the NAS(GGSN) sends an accounting-on request after a reload, the SSG will clear the existing Host Object. It will take some time to clear (depends on the number of host objects and the accounting stop rate-limit that was configured). During that time period, the SSG ignores the new create request from that GGSN.

Workaround: This behavior can be resolved by reloading the NAS(GGSN) with 20,000 users logged in to SSG with the accounting stop rate-limit configured at 100. As a result, the SSG will accept a new create request after approximately 2-3 minutes.

Send 20,000 RADIUS sessions at a rate of 10 ps with one service for each session using the RADIUS client.

Send account-on from the RADIUS client with the same NAS-IP as used in step a.

Immediately send a further 20,000 RADIUS sessions from the same NAS client at a rate of 10 ps.

Resolved MWAM Caveats

Open Caveats—Cisco IOS Release 12.3(5a)B1

This section documents possible unexpected behavior by Cisco IOS Release 12.3(5a)B1 and describes only severity 1 and 2 caveats and select severity 3 caveats.

Open SSG Caveats

When downstream traffic from the service network is sent as multiple packets to the same connection object before sending packets to the next connection object, processor usage is much less than if packets are sent consecutively to different connection objects (even though the rate of traffic sent for both the cases is the same).

Open MWAM Caveats

On rare occasions, the MWAM displays a minor error in response to the show mod command after the Supervisors switch activity for a failover condition:

Resolved Caveats—Cisco IOS Release 12.3(5a)B1

All the caveats listed in this section are resolved in Cisco IOS Release 12.3(5a)B1. This section describes only severity 1 and 2 caveats and select severity 3 caveats.

Resolved SSG Caveats

A router with active SSG tunnel services and user logging into and logging off tunnel services may see an increase in the memory usage.

An SSG with permanent tcp-redirect (plug-and-play) enabled may reboot when the user connects it to a service with the following attribute: ssg-service-info "KW0" (which means "no permanent http redirect for this service").

Some of the user access-request messages may be dropped by the SSG when the down-stream client sends an accounting-off message.

When the SSG forwards accounting requests from the NAS to the AAA server and one packet gets lost between the SSG and AAA, the SSG behaves incorrectly.

Despite configuring PZI0 in the SSG service profile, Interim Accounting Records are regularly sent to the prepaid server, which is configured locally on the SSG router.

–

Pass-through service #1—Pass-through service with only single hosts with ACL

–

Pass-through service #2—Pass-through service with single hosts and some networks with corresponding ACL

Packets from hosts that are not in pass-through service #1 or pass-through service #2 and that are sent to the subscriber IP address are passed to the subscriber through the tunnel interface. The reply packet from the subscriber is then sent through the tunnel. Both packets are billed in the tunnel service.

The PZW attribute, when configured in the service profile, sends prepaid authorization for the user only when traffic is received from the user. But the connection accounting start is sent after the user is logged into the service. Since the prepaid server has not performed authorization for the user, receiving the accounting-start message confuses the server.

When a multilink interface is configured as an SSG downlink interface, the ssg direction downlink command is removed from the interface configuration when the multilink interface flaps.

When the CIR is configured for both upstream and downstream and the Normal Burst size and Excess Burst size are not configured in the service profile, the SSG displays the wrong value for the Excess burst rate in the connection object.

The SSG crashes when permanent redirection is configured after Captive Portal has already sent a message to the SSG that a user has Web proxy settings.

The ppp timeout idle seconds command configured under Virtual-Template does not work on the LAC for VPDN sessions forwarded to the LNS.

While attempting to create 65000 host objects with two services, the SSG crashed after creating 40000 host objects.

SSG with TAL configuration does not use all parameters provided by AAA server in the Access Accept. This is required when SSG needs to create a host object (HO) that is deleted after timeout.

Resolved MWAM Caveats

The processor control (PC) in the MWAM may become unreachable when packets are sent through one of the Sibyte processors at a very high rate.

In rare cases, the PC may freeze without any error message on the console. There are no keepalive messages at the processors, and it is not possible to session to the any of them.

During an RPR+ switchover on a dual Supervisor chassis, the RPR+ operation can stall when the SFM-capable 48-port 10/100 Mbps RJ45 linecard (WS-X6548-RJ-45) fails to go on line. The linecard is automatically powered down, and the RPR+ process does not execute a timely switchover to the second Supervisor. The system can be off line for up to five minutes before recovering. The linecard recovers and powers on without operator intervention.

The user can observe the following failure messages (module 9 is the WS-X6548-RJ-45 card):

%OIR-SP-3-PWRCYCLE: Card in module 9, is being power-cycled off

          (Module not responding to Keep Alive polling)

SP: oir_notify_online: Failed to send online notification: slot 9

%DIAG-SP-6-DIAG_OK: Module 9: Passed Online Diagnostics

%OIR-SP-6-INSCARD: Card inserted in slot 9, interfaces are now on line

On rare occasions after an MWAM is reset, there is no IP connectivity between the MWAM processors and the Supervisor or any outside devices. This problem may occur after issuing the hw-module module <#> reset command from the Supervisor.

Traffic flowing from an MWAM interface into a VLAN exceeds the policing limits defined by the policy applied to the VLAN.

When a MWAM internal interface has been defined to be part of a specific VLAN through the mwam module <module-num> port <port-num> allowed-vlan command, and the VLAN has a QoS policy applied to the input which contains a police policy, the traffic coming from the MWAM will exceed this policing definition.

The MWAM configuration mode is switched from supervisor mode to local mode after the MWAM is reloaded. This occurs when the radius-server unique-ident <#> command is configured on MWAM processors.

Resolved Caveats—Cisco IOS Release 12.3(5a)B

All the caveats listed in this section are resolved in Cisco IOS Release 12.3(5a)B. This section describes only severity 1 and 2 caveats and select severity 3 caveats.

A vulnerability in the Transmission Control Protocol (TCP) specification (RFC793) has been discovered by an external researcher. The successful exploitation enables an adversary to reset any established TCP connection in a much shorter time than was previously discussed publicly. Depending on the application, the connection may get automatically re-established. In other cases, a user will have to repeat the action (for example, open a new Telnet or SSH session). Depending upon the attacked protocol, a successful attack may have additional consequences beyond terminated connection which must be considered. This attack vector is only applicable to the sessions which are terminating on a device (such as a router, switch, or computer) and not to the sessions that are only passing through the device (for example, transit traffic that is being routed by a router). In addition, this attack vector does not directly compromise data integrity or confidentiality.

All Cisco products which contain TCP stack are susceptible to this vulnerability.

and it describes this vulnerability as it applies to Cisco products that run Cisco IOS� software.

A companion advisory that describes this vulnerability for products that do not run Cisco IOS software is available at:

All Cisco products which contain TCP stack are susceptible to this vulnerability.

and it describes this vulnerability as it applies to Cisco products that run Cisco IOS� software.

A companion advisory that describes this vulnerability for products that do not run Cisco IOS software is available at:

Resolved SSG Caveats

There is a drift of time in the interim accounting records generation over a period when compared to the configured interval.

This issue occurs during high traffic, saving config or causing a nvram write can cause packets to be dropped.

Workaround: Lower traffic rate when performing maintenance such as configuration file saves.

	%ALIGN-1-FATAL: Corrupted program counter

	 pc=0x0, ra=0xXXXXXXXX, sp=0xXXXXXXXX

	%ALIGN-1-FATAL: Corrupted program counter

 	pc=0x0, ra=0xXXXXXXXX, sp=0xXXXXXXXX

	Unexpected exception, CPU signal 10, PC = 0x0

A timer wheel may fail when the same timer is started from both the process level and the interrupt level.

This issue is observed on a Cisco router that runs Network Address Translation (NAT).

This issue is observed on a Cisco router that receives excessive multicast control traffic.

Workaround: Apply a quality of service (QoS) policy map to limit the rate of the multicast control traffic that can be received by the router.

You may not be able to configure the maximum transmission unit (MTU) on a virtual template.

The default number of missed keepalives required to bring down a ppp link has changed from 5 to 3 in releases that have integrated CSCdt94888. The original default behavior can be restored by configuring "keepalive 10 5"on the interface.

The service selection gateway (SSG) sends duplicate Acct-Session-Id in the SSG connection accounting record. Same session id is used in the user accounting record.

When VPDN session is disconnected by authentication failure, no VPDN syslog message (%VPDN-6-AUTHENFAIL) and history failure table are logged. A record is overwritten by normal causes (%VPDN-6-CLOSED, Result 1, Error 0)

When the ip radius source-interface global configuration command is configured on a PPP over Ethernet (PPPoE) server, the interface address may not be set in the RADIUS NAS-IP-Address [4] attribute.

This issue is observed on a Cisco platform that runs Cisco IOS Release 12.3(2), 12.3(2)T, 12.3(3)B, or 12.3(4)T, that functions as a PPPoE server, and that has the radius-server attribute nas-port format format global configuration command enabled with the value d for the format argument.

Workaround: Do not use value d for the format argument. Rather, use another value to configure the network access server (NAS) port.

Alternate Workaround: Enter the radius-server attribute 4 nrp global configuration command.

The individual AAA periodic accounting update messages (Radius accounting messages with Acct-Status-Type=Watchdog) generated by an IOS gateway for each call leg (TDM and IP) of the same voice call may be sent to the Radius server more than 5 minutes apart due to the randomized timer algorithm used by the AAA message transmit function.

When ip address negotiate is configured on an interface and our address is not successfully negotiated with the peer, no address is assigned to our interface which can cause problems with IP/CEF forwarding.

The memory held by SSGCmdQueue process increases continuously when SESM users log on and log off.

Without any global radius servers configured, an access-request is sent to the server defined in the AAA test server group. This happens even with no "radius-server key" defined. This behavior does not occur is 12.2(13.7)T, the error message "No radius servers defined" is displayed.

This is not a serious issue and is a configuration problem. The user is warned when a server that has not been defined is added to the server group.

router(config)#aaa group server radius bogus

router(config-sg-radius)#server 10.1.1.1 ?

  acct-port  UDP port for RADIUS accounting server (default is 1646)

  auth-port  UDP port for RADIUS authentication server (default is 1645)

router(config-sg-radius)#server 10.1.1.1

00:55:48: %RADIUS-4-NOSERV: Warning: Server 10.1.1.1:1645,1646 is not defined.

It is expected that the behavior will be undefined if the user does not correct the misconfiguration.

SSG may not send a calling station ID in connection accounting records to a local and a remote AAA server.

This issue is observed when a client log on by using a proxy service with MSISDN.

Time Drift in Interim Accounting update was seen for SSG connection accounting packets.

This issue occurs with 10 Host Objects and Connection Accounting interval 300. After 4 days of testing time drift was seen in Interim accounting update packets.

This issue is observed on a Cisco router when CEF or PXF is enabled and when the encapsulation is changed while no VC is defined.

Currently for prepaid services, the initial authorization to obtain quota happens during the time a user is logged on to a service. This fix enables prepaid services to be configured so that the initial authorization happens only when the user starts sending traffic on that connection.

The attribute to be configured in prepaid services for this purpose is shown below:

	a = 26  (Radius attr for vendor specific)

	b = len (length of the Radius Vendor specific Attribute>

	d = 251 (Sub attribute ID for Service-Info)

	e = len (length of the vendor specific sub attribute)

The accounting start would be sent during activation and not after user traffic.

When a new image is loaded on an MWAM, if either of the processors of a complex previously had configurations in the startup-config that are not recognized by the new image, then the complex may keep resetting continually.

Workaround: If a new image to be loaded on the MWAM may not have one or more subsystems of the previous image and may not recognize a large number of configurations in the startup-config, make sure those configurations are removed from the processors (unconfigured and saved) before loading a new image.

Resolved MWAM Caveats

MWAM VLAN interfaces stop responding when the Cisco 7609 router is rebooted. Ping packets sent from the Supervisor to the MWAM fail.

SNMP query for CISCO-FLASH-MIB does not populate values. The fields of the CISCO-FLASH-MIB are currently not populated for the flash devices dedicated to each of the processors of the MWAM. When the CISCO-FLASH-MIB of a MWAM processor is queried, the fields of this MIB will incorrectly appear as if there is no flash device for this processor.

When multiple MWAMs are reset at the same time using the hw-module module slot_number reset command, on rare occasions the MWAM will fail to boot (remain in a PwrDown state) and the following message will display on the Supervisor console:

SP: oir_disable_notice: slot12: lcp failed to go online

After reloading a switch containing ten MWAMs, it is no longer possible to upgrade the MP or AP images. All attempts fail with the following message:

stress-6500a#copy tftp: pclc#6-fs: Address or name of remote host [64.102.16.25]? 
Source filename [users/gferris/mwam/c6svcmwam-js-mz.geo_t_030924.1-2-2-1.dev.bin]? 
Destination filename 
[users/gferris/mwam/c6svcmwam-js-mz.geo_t_030924.1-2-2-1.dev.bin]? Accessing 
tftp://64.102.16.25/users/gferris/mwam/c6svcmwam-js-mz.geo_t_030924.1-2-2-1.dev.bin... 
Loading users/gferris/mwam/c6svcmwam-js-mz.geo_t_030924.1-2-2-1.dev.bin from 
64.102.16.25 (via Vlan111): ! %Error opening 
pclc#6-fs:users/gferris/mwam/c6svcmwam-js-mz.geo_t_030924.1-2-2-1.dev.bin (Error 
Sending Request) stress-6500a#

When simultaneous hw-module reset commands are issued from the Supervisor, some of the MWAMs do not respond correctly. When the reset is issued, the following message displays:

Oct 1 01:49:20: SP: The PC in slot 7 is shutting down. Please wait ... If the problem 
occurs, the following messages are displayed: Oct 1 01:52:20: SP: 
shutdown_pc_process:No response from module 7 Oct 1 01:52:20: %C6KPWR-SP-4-DISABLED: 
power to module in slot 7 set off (Reset) *Oct 1 01:52:19: 
%C6KPWR-SP-STDBY-4-DISABLED: power to module in slot 7 set off (Reset)

Under certain circumstances a processor may reload when deleting a file from the bootflash partition.

Any debug messages between the MWAM processors and the PC that do displays on the processor console, also do not appear in the Remote Console and Logging (RCaL) debug. For example, the heart beat debug message from the processors to the PC do not appear in the RCaL debug.

Global Reset: Unable to Initialize BootManager: Can't open device

The show log upgrade command displays the upgrade log of an AP upgrade when the user is in the MP. However the command does not display the upgrade log of an MP upgrade when the user is in the AP.

The show processor command on the PC should differentiate IOS reloads from the PC, user reloads, and unexpected reloads.

The MWAM cannot be shut down or reloaded. If a reload is issued, the module is eventually reset by the Supervisor; if a shutdown is issued, the module remains in the other state indefinitely.

After a Supervisor switchover, one or more MWAM processors fails to become active.

When NTP is used in MWAM processors for time sync, the time sync is lost with the NTP server on an MWAM reload. Individual processor reloading does not cause any problem.

Workaround: Remove and reconfigure NTP commands from the running configuration on all the processors.

Resolved Caveats—Cisco IOS Release 12.3(3)B1

All the caveats listed in this section are resolved in Cisco IOS Release 12.3(3)B1. This section describes only severity 1 and 2 caveats and select severity 3 caveats.

Resolved SSG Caveats

SSG system shows tracebacks and reloads with unexpected exception, CPU signal 10, PC = 0x613F1C10.

This problem occurs when a user is cleared by CLI or disconnects by switching CPE off. Exact cause not yet known.

This condition was caused by a password mismatch between SESM and SSG in the test setup.

This problem occurs when SSG does not send re-authorization request after Quota Time expiry for connection with QT60, QV0 and Idle 0.

This condition was caused by a password mismatch between SESM and SSG in the test setup.

This problem occurs when SSG trys to allocate memory for proxy service authorization packet.

The SSG hangs (no console or Telnet access) after data traffic is started. The SSG will not reload and must be power-cycled.

With PZI60 and L60 in the service profile, the SSG sends Interim accounting updates alternatively to local and prepaid server.

A Cisco router with the SSG application may reload. This is noticed with Cisco MSID access request and when the access accept from a AAA is delayed and/or the access response does not contain CDMA Realm.

A Cisco router that terminates both PPP over Ethernet (PPPoE) and PPP over ATM (PPPoA) sessions may fail to switch traffic downstream toward the subscriber via Cisco Express Forwarding (CEF) for a period of up to three minutes.

This symptom is observed when the PPPoE and PPPoA sessions use different virtual templates and when subinterfaces are enabled. The symptom may affect only some subscribers.

The Calling-Station-Id is not sent in connection accounting records. This condition occurs for proxy service logon with MSISDN.

The wrong Calling-Station-Id is sent to the LNS during tunnel service creation. When a different Calling-Station-Id is received from the SESM for tunnel service logon, the SSG should use this Calling-Id for tunnel service creation with the LNS. However, the SSG is incorrectly sending the host logon calling-id to the LNS for tunnel creation.

The RADIUS attributes that contain the CALLING and CALLED numbers are not in the service account records.

When an HSRP SNMP query is performed on a router with an HSRP group configured on a subinterface, the router stops responding and eventually reloads. This action does not occur for HSRP groups configured on major interfaces.

When ip radius source-interface is configured on the PPP over Ethernet (PPPoE) server, the interface address is not set in the RADIUS NAS-IP-Address [4] attribute.

Workaround: Configure the command radius attribute 4 nrp. the problem can be circumvented. Hence not a show stopper.

If the number of sessions exceeds the configured session limit on the L2TP network server (LNS), subsequent session requests cause a memory leak in the L2TP management daemon.

Original problem (CSCeb65615): The PPP idle timer on a virtual access interface resets with uninteresting outbound traffic that is defined with the command ip idle-group <acl> out. The ACL that defines the uninteresting traffic finds no matches even though the output traffic is uninteresting. This problem was partially repaired in CSCeb65615. Remaining restrictions are resolved with CSCeb84730. These restrictions are:

–

Virtual access subinterfaces should not be used with idle-group configuration because, under some conditions, the idle timer is not reset even by interesting traffic.

–

If the virtual access interface has an inbound ip access-group where the ACL has log keyword, the idle-group inbound ACL shows no matches even though it works.

The PPP timeout AAA inbound does not prevent outbound packets from resetting a per-user idle timer. This problem was partially resolved with CSCeb82500. Remaining problems are resolved with CSCec10191.

The SSG upstream counter statistics for the connection object are incorrect. This condition is observed when the SSG TCP redirect feature is enabled.

Workaround: Disable IP CEF (no ip-cache cef) on the downlink interface. Note: This action reduces packet throughput.

Accounting On packets that are sent by an Access Zone Router (AZR) that has had a cold restart (that is, power is turned off and then on again) may not be acknowledged by an SSG. This symptom is observed on an SSG that is configured with a basic RADIUS proxy setting.

Workaround: Create a new RADIUS group, and configure the SSG to forward all accounting messages to this new RADIUS group.

Resolved MWAM Caveats

MWAM processors are reloaded when receiving traffic at 100% processor capacity. The condition occurs when sending downstream traffic on processors configured in a cluster.

The IOS on MWAM processors does not support NTP and other clock commands. The processors synchronize their clocks from the PC, which in turn synchronizes with the Supervisor clock. The Supervisor clock is linked to an NTP clock source.

When the clock value is displayed with show clock command both on the Supervisor and an MWAM processor, the values between the Supervisor and the processor clocks is off by a couple of seconds.

The MWAM processor/complex may reload when many VRFs are being configured/unconfigured and data flows to these networks are on.

If the ROMMON needs upgrading during configuration of the Remote Console and Logging (RCaL) on the MWAM, the user should be notified through a console message.

Resolved Caveats—Cisco IOS Release 12.3(3)B

All the caveats listed in this section are resolved in Cisco IOS Release 12.3(3)B. This section describes only severity 1 and 2 caveats and select severity 3 caveats.

Resolved SSG Caveats

SSG makes authorization requests towards a prepaid server even though the connection cannot be activated.

SSG makes a service authorization request towards OCS for a prepaid service, before it checks whether this service can be activated or not. The service authorization request causes the OCS (prepaid server) to deliver a quota, but if the quota cannot be used by the SSG, this unused quota will not be returned to OCS for other active services. One reason why a service cannot be activated could be that the service is pointing to the same network as another service.

Workaround: Mark services with overlapping service networks as sequential or part of the a mutually exclusive service group so that user cannot log into both of them simultaneously.

A Cisco router that has a Quality of Service (QoS) service policy attached to an interface may generate memory alignment errors or reload unexpectedly because of a bus error during normal mode of operation.

This problem is observed when the policy map of the service policy has a set action configuration and when traffic is being processed.

When VPDN session is terminated by PPP authentication failure, no VPDN syslog message (%VPDN-6-AUTHENFAIL) and history failure table are logged.

SSG forwards accounting retransmits from radius-clients to the AAA server and also does additional retransmits for each forwarded request.

When SSG is configured to forward accounting requests from radius-clients, if the AAA server responds slowly, radius clients will retransmit the accounting requests. SSG forwards the accounting requests even though it is waiting for a response from the AAA server. However, for access-requests from radius-clients, SSG does not forward retransmitted access-requests while waiting for a response from AAA server.

Workaround is to make the radius-client (Radius timeout*Retry) time greater than SSG (Radius timeout*Retry) time.

The bug was reported on Catalyst 6000. Other platforms also may have this issue. In hybrid mode, when a reset [15/16] is issued from SP (CatOS, hybrid mode), the reload time displayed is very large.

Enhanced Interior Gateway Routing Protocol (EIGRP) hello messages may be sent from a virtual-access interface when they should not be sent.

This problem is observed on a Cisco router that has the passive-interface default or passive-interface virtual-template interface-number router configuration command enabled.

This problem occurs if the primary service logon fails because of authentication. Any wrong tunnel parameters in the tunnel profile can crash the box.

Configure the router to send accounting start and stop records for a exec connection and configure the following command: aaa accounting send stop-record authen fail.

Do a Telnet to the router from any other router. Do not enter anything when it prompts to enter a username. After some time it times out and will say "[Connection to <IP Addr> closed by foreign host]"

When the Telnet connection times out, two accounting stop records are generated.

RADIUS server is marked dead and does not show as "UP" after the deadtime interval has expired.

Two RADIUS servers are configured on LNS, one of them is marked as Dead during the bootup process because it was not able to respond to system accounting request.

When PPP sessions come up, LNS is still trying to send RADIUS request to dead RADIUS server but now it can access that AAA server because LNS builds up the routing information. LNS is getting responses back from the RADIUS server.

LNS is not changing the status of that RADIUS server to UP even after the elapse of configured dead time.

When SSG control error debugs are enabled, "Stale network routes" error message is displayed.

This will happen if there are exclude networks ("E") configured in the service profile and if the user logon to this service and does a account logoff.

In Cisco IOS 12.3 B releases with CSCeb30098 integrated, LCP renegotiation at the L2TP Network Server after authentication has already completed will cause the session to enter the wt-sss state (as seen in "show vpdn"). Unless the LAC tears down the session, the session may get stuck in the wt-sss state.

The set commands that are used with a service policy can cause a router to reload in some circumstances. The set cos policy-map class configuration command can cause reloads in addition to other set commands.

This problem may be observed with configurations that have a service policy with the set command on the interface.

User does not get connection to service, for a PPPoE user when logs in second time.

This problem is seen with the PBHK enabled and the PPP session is created as a non SSG PPP user session.

Also the show ssg host command shows an error message that prints that memory is low. The show ssg host count shows that host count is -ve.

1. SSG binds the PPPoX interface dynamically as downlink (because "ssg direction downlink" has been configured under virtual-template interface mode)

SSG host is not deleted when the PPP session goes down. Also when the host is deleted using "clear ssg host all", the host count becomes -ve.

Work-Around: Make sure that the condition#1 does not occur. This can be done by inserting a dummy ssg-account-info attribute in the access-accept of the PPPoX user. This dummy attribute can be: ssg-account-info "Nabracadabra"

A Cisco router terminating both PPPoE and PPPoA sessions may fail to CEF switch traffic downstream toward the user when different vtemplates are used for the two types of sessions and sub-interfaces are enabled. This problem may affect only a portion of the subscribers.

Workaround: Use one vtemplate for both types of sessions, disable sub-interfaces or disable CEF.

The input queue of the Gi0/0 interface on MWAM module, used by a sibyte processor running the SSG application, becomes full if a AAA server failure occurs. From that point on, no traffic is forwarded between the MSFC and the subinterfaces configured on Gi0/0 from within the SSG application on the sibyte (pings between MSFC and subinterfaces on Gi0/0 fail, and so on).

 %AAA-3-BADMETHODERROR: Cannot process authorization method SERVER_GROUP

 %AAA-3-BADMETHODERROR: Cannot process accounting method SERVER_GROUP

-Process= "AAA Server", ipl= XXX, pid= YYY

where XXX and YYY are arbitrary integers greater than or equal to zero. The router then produces a traceback.

This problem is observed when you configure and then attempt to use an authorization or accounting method list which refers to a server group which contains no servers, and which has never contained any servers since the router booted.

aaa authorization network default group radius but did not configure any RADIUS servers globally, you would see the error message every time a user attempted to perform network authorization.

Workaround: Make sure that the server group contains at least one server. To add a RADIUS server to the global group "radius", configure:

PPPoE sessions does not come up when some debugs are enabled in the LAC. This could possibly due to the additional time lag introduced by enabling the debugs in the LAC.

This will not happen when "lcp re-negotiation" is not configured in the virtual-template in the LNS side.

SSG misbehaves (and often crashes) after total number of connections on the box become 64K.

When a user is connected to a service with certain networks, upstream packets from user towards service will be dropped.

SSG-DATA: CEF-UPST: Unable to find adjacency. Punt (FastEthernet0/0 : 
10.0.1.1->10.1.1.1)

SSG-DATA: PROC-UPST : IDB is NULL. Drop (FastEthernet0/0 : 10.0.1.1->10.1.1.1)

This happens when the destination address falls into a service network of

Workaround: Replace the service network so that at least one bit matches the destination address.

While using SSG, executing show align< indicates that a spurious memory access has occurred.

AVP 31 (Calling-station-id) is missing from accounting records to prepaid server when SSG radius-proxy users are accessing prepaid service. It happens only when no explicit calling station id is available to SSG.

This problem was first reported on cisco 7200 platform but same exists on all Cisco platforms supporting SSG functionality.

The accounting of input and output bytes/packets for a service connection is not correct. Only upstream traffic is accounted for that service access whereas downstream traffic from that service would be accounted for another service connection.

Downstream traffic to a ssg host logged onto a proxy NATed service. This happens after a host logs off a service and immediately same/another host with same NATed IP address logs on to the proxy NATed service.

Mem-leak is seen when SSG subscriber access his SOHO and the user is logged on to a Tunnel service.

For each authorization retry in timeout quota in SSG traceback at ServiceAuthorize() is seen.

This can happen if the box is running out of memory and TCP-Redirect is configured.

Real IP assigned by service for an ssg connection is sent as framed-ip attribute in the access-accept to SESM.

When a service (proxy or tunnel) assings an IP Address for a connection SSG send it to the SESM in response to the service logon request in the framed-ip attribute. This hides the framed-ip of the host in the access-accept.

Any new Access Requests from NAS(GGSN) are not processed by SSG when SSG_dummy_pool fills up.

SSG_dummy_pool fills up when SSG is honoring an Acct-on/Accounting Off along with an accounting stop throttle configuration. Any new Access-Requests from NAS(GGSN) can create this condition.

Workaround: Unconfig and config "ssg radius-proxy" OR a Reload of SSG will clean up this pool.

If the Connection to the LNS fails (due to LNS Reboot or redundant LNS-Failover) the SSG needs a long time to send L2TP HELLO packets to tear down control connection and re-establish tunnel to redundant LNS. During this period several L2TP-HELLOs are sent to the LNS.

The SSG supports vpdn-group names in the SSG tunnel service profile. The L2TP-specific configuration (for example, tunnel gateway address, tunnel password, and so on,) can be configured in the vpdn-group on the device; only the group name needs to be specified in the service profile.

The SSG uses this vpdn-group name in the service-profile to retrieve the VPDN configuration and set up the tunnel session. This enables the SSG to support all L2TP tunnel parameters configurable within the VPDN group command for setting up tunnel sessions.

The service-profile for the tunnel service accepts the following Cisco Generic VSA:

cisco-generic = "vpdn:group-name=<name>"

    cisco-generic = "vpdn:group-name=tunnel_corp"

    ssg-service-info = "R10.0.0.0;255.255.0.0"

Resolved MWAM Caveats

Workaround: Use the show version command to view the IOS image from the MWAM processor.

An attempt to copy a file to the bootflash:partition of an MWAM processor with a destination filename that already exists on this partition will fail. A copy cannot be made to a file that already exists. The following error message is displayed:

MWAM traffic shaping does not function with MWAM Gigabit Ethernet interfaces. Traffic shaping configurations on MWAM gig0/0 interface has no effect. The driver for MWAM gig0/0 interface does not support traffic shaping.

When an MWAM is removed from a slot, the MWAM configuration files remain with the MWAM. A replacement MWAM in the same slot must then be fully reconfigured. Also, when an MWAM is moved from one slot to another, the configuration files move with the MWAM instead of being associated with the original slot.

Whenever you perform the copy running-config startup-config or write memory operation from an MWAM console, always use the copy startup-config tftp://server_name/file_name to copy the MWAM configuration file to an external server. Perform this operation for each MWAM processor.

Before moving the MWAM, issue the write erase command at the console of each MWAM image.

After installing the MWAM in its new slot, issue the following commands at the consoles of each MWAM processor:

copy tftp://server_name/file_name running-config

copy running-config startup-config

Note

If a TFTP server is unavailable, any bootflash device (slot0: or disk0:) on the Supervisor module can store the MWAM configuration files. This alternative requires configuring the Supervisor for RCP only (not TFTP). It also requires creating empty (that is, dummy) configuration files on the Supervisor module. The MWAM configuration files are addressed to the Supervisor module using the address:128.0.0.x where x is the Supervisor slot (for example, 128.0.0.1). The bootflash then becomes the preferred device. Use the file naming convention SLOTxPCy.cfg, where x is the MWAM slot and y is the MWAM processor number. This convention facilitates migration to a future MWAM feature that resolves this problem.

SNMP query for variable chassisType(1.3.6.1.4.1.9.3.6.1) returns -1 for MWAM module.

Issuing the copy running-config startup-config command from the MWAM console fails to write the configuration to the standby Supervisor module.

Issue the mwam bootflash access command from the Supervisor console. If a switch-over occurs or if you reload the standby Supervisor module, you must re-issue the mwam bootflash access command.

Generate the startup-config file, copy it (TFTP) to a location for editing, and add the line mwam bootflash access. Then copy the file back to startup-config and reload the Supervisor modules. This action enables mwam bootflash access on reloading. However, if you copy the running-config to startup-config on the Supervisor, you remove this configuration and must repeat this workaround.

When a chassis is reloaded and contains multiple MWAMs that are running in the Supervisor configuration mode (that is, MWAM configurations stored on the Supervisor bootflash), some of MWAM processors may not receive their configurations from the Supervisor bootflash.

Reset the MWAM from the Supervisor console and verify the configuration on each processor. If a processor is found to have no configuration file, reload only that processor.

Use local configuration mode on the MWAM instead of the Supervisor configuration mode.

MIBs

To obtain lists of supported MIBs by platform and Cisco IOS release, and to download MIB modules, go to the Cisco MIB website on Cisco.com at the following URL:

Table Of Contents