Table Of Contents
Release Notes for Cisco Service Control Operating System (SCOS) 3.1.6
Link loss time of 40 milli in power boot when OPB is active
DHCP and RADIUS flows were disrupted by the ICCF sanity-check mechanism
Packets should not be injected into sprayed flows in asymmetric tunneling
Packet injection to initiator side, into flows with no initiator payload, in asymmetric tunneling
'no service telnetd' command did not block the Telnet port
SCE2000 automatically reloaded due to a rare internal error in the DSCP marking logic
Potential memory overrun in high availability configuration with a high number of subscribers
Invalid packet generated in MPLS traffic engineering mode with DSCP marking
DSCP marking+inject+tunneled traffic generated malformed injected packet
`show snmp MIB pcube-SE-MIB port' returned wrong number of ports
Applying a policy failed when the SCE was in memory congestion
Flow filter counters in 'show linecard counters' command were not meaningful to user
Simple CLI command needed for globally setting asymmetric Layer 2 support
FTP password was visible with show version and package extract
Identifying Subscribers per VLAN ID and IP
Flexible Subscriber Introduction
Upgrading the SM caused SCE failover in cascade topology
Non-first fragments misclassified
`debug' command caused reboot with fatal error
ip ftp username and password did not work
Upgrading to PP#11 caused error
Defining traffic rules not permitted at the admin authorization level
SCE platform should block simultaneous `get support-file' commands from multiple interfaces
TCP Learning did not close flows that started with SYN-ACK
Misclassification in MPLS/VPN auto-learn mode
Using TIR in MPLS environment caused SCOS - CAM crash
First subscriber had no expiration time (3.1.5LA)
SCE-Sniffer RADIUS LEG did not work after PQI installation
Subscribers data synchronization was slower when static subscribers were configured
CLI command `IP address' did not indicate to the user that a reload of the SCE is required
RADIUS/DHCP sniffer in SCE might stop functioning for certain flows
SCE might reload in rare cases due to an infinite loop in its scheduler
Problem with fragments classification consumes log assets
Global-Controllers triggers warnings to the log when querying the MIB Group
Bad response to a SNMP query and a Warning in DBG log
Uni-directional classification for support of asymmetric routing
Subscriber synchronization in cascade setups
SSH activity sometimes ended in an unstable state
Redirect packet was sometimes transmitted on wrong link
Dynamic Signature (DSS) caused a reboot after dozens of consecutive Apply operations
Reload of the SCE due to an expiry of an internal watchdog mechanism
The saving process of application configuration was not resilient to reload
Failure to retrieve support file using FTP
PRPC authentication security level was set to default after RPC adapter restart
Part of the quota information was not exchanged between two cascaded SCE platforms
Subscriber with many mappings did not send all lease time expiration notifications
Link reflection failed to operate after long uptime
Attack-detector port-list showed random numbers in running-config
Potential discard of packet under extreme network conditions
Access violation when configuring anonymous groups
rdrActiveConnectionTrap was not sent after reload of the SCE platform
Link up/down traps were not sent on all ports
MIB object 'pmoduleType' returned wrong value in cascade setups
MIB variable ifMtu returned wrong value for traffic ports
Some rdr-formatter MIB objects returned information only for category 1
SCE reloaded due to watchdog expiry
Refinement of the maximum number of IP mappings per subscribers
Unused ACLs disappear after applying policy
When L2TP skip is configured, non L2TP fragments may cause flow mismatch
Quota events not received by SCE subscriber API client or QM
While removing all VPNs management operations cannot be performed
Tunnel-id-based traffic rule applies DSCP marking to non-tunneled traffic
SCE2000 may reload if stalled by flow control
SCE2000 may drop cascade control packets from AF4 queue during GC congestion
Link failure may be reflected to all ports if a port is flickering due to a HW problem
The configured attack threshold is set for each PPC separately
Flow `opened from VAS' is misrouted if there is a FF rule to bypass
Packet Loss during Application Installation or Upgrade
Obtaining Documentation and Submitting a Service Request
Release Notes for Cisco Service Control Operating System (SCOS) 3.1.6
Revised: August 10, 2009, OL-14438-15Introduction
These release notes for the Cisco Service Control Operating System describe the functional enhancements and fixes provided in Cisco Release SCOS 3.1.6. These release notes are updated as needed.
For information regarding features added and issues resolved in the 3.0.x train, please refer to Release Notes for Cisco Service Control Operation System (SCOS) 3.0.6.
For a list of the caveats that apply to Cisco Release SCOS 3.1.6 see Open Caveats.
Supports: SCOS 3.1.6, SCOS 3.1.5, SCOS 3.1.1, and SCOS 3.1.0.
•
Obtaining Documentation and Submitting a Service Request
Introduction
Cisco is proud to release version 3.1.6 of the SCOS (Service Control Operating System) for its SCE platform.
SCOS 3.1.6 is a maintenance release of SCOS 3.1.0. It includes fixes of issues that were identified as part of Cisco's on going internal testing and during our interaction with our customers.
This document outlines the resolved issues delivered in the SCOS 3.1.6 release. It assumes the reader already has a good working knowledge of the Cisco Service Control solution. For additional information, please refer to the Cisco Service Control Engine documentation.
Note
1.
2.
3.
4.
SCOS RELEASE 3.1.6
Compatibility Information
SCOS 3.1.6 may be installed on the following Service Control Engine platforms:
•
•
•
SCOS 3.1.6 is not compatible with the following Service Control Engine platform:
•
Functional Enhancements
The following section lists the functional enhancements in SCOS release 3.1.6. See the Cisco Service Control Engine Software Configuration Guide for more information regarding these features.
IPinIP
IPinIP is an IP-based tunneling protocol in which the system must be specifically configured to recognize the flows inside the tunnel. The SCE platform will then skip the external IP header, reaching the internal IP, which is the actual subscriber traffic.
For IPinIP traffic, DSCP marking can be configured to be done on either the external or the internal IP header exclusively.
See the IPinIPsection in the Cisco Service Control Engine Software Configuration Guide for more information.
Resolved Issues
The following issues were resolved in this release.
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Link loss time of 40 milli in power boot when OPB is active
•
Power failure should activate the optical bypass immediately. In practice, switchover time can be about 40 milliseconds, depending on the exact hardware or firmware version in the system.
This issue is fixed in SCOS 3.1.6.
DHCP and RADIUS flows were disrupted by the ICCF sanity-check mechanism
•
When the ICCF sanity check mechanism was enabled, it sometimes interfered with business critical flows (such as DHCP or RADIUS), by placing them in bypass. This in turn stopped RDR generation.
This issue is fixed in SCOS 3.1.6.
Packets should not be injected into sprayed flows in asymmetric tunneling
•
Permitting sprayed flows to inject packets when in asymmetric tunneling modes made it possible that L2 switches might be confused by the source MAC address, causing incorrect MAC learning in the connected layer 2 switches.
This issue is fixed in SCOS 3.1.6.
Packet injection to initiator side, into flows with no initiator payload, in asymmetric tunneling
•
In asymmetric L2 scenario, when requested to block a flow on which the SW had not seen packets from the initiating side, the SCE used the L2 information in the current packet instead of avoiding the inject, possibly disrupting MAC learning on the switches.
This issue is fixed in SCOS 3.1.6.
Packet injection to non-initiator side, into flows where non-initiating side info is unknown, in asymmetric tunneling
•
In asymmetric L2 scenario, when requested to block a flow on which the SW had not seen packets from the non-initiating side, the SCE used the L2 information in the current packet instead of avoiding the inject, possibly disrupting MAC learning on the switches.
This issue is fixed in SCOS 3.1.6.
'no service telnetd' command did not block the Telnet port
•
Disabling the Telnet server of the SCE (using the CLI command 'no service telnetd') disables any new Telnet connection to the SCE platform, but did not block the Telnet port.
This issue is fixed in SCOS 3.1.6.
Enabling/Disabling SSHv1
•
When the SSH server was enabled, there was no mechanism to enable only SSHv2 (the most recent version of SSH).
This issue is fixed in SCOS 3.1.6. You can now disable SSHv1 so that only SSHv2 is running.
SCE2000 automatically reloaded due to a rare internal error in the DSCP marking logic
•
The combination of an internal error in the DSCP marking logic in the SCOS together with a long processing time of a single packet could have resulted in a false expiry of the SCE watchdog mechanism and reload of the SCE platform.
This issue is fixed in SCOS 3.1.6.
Potential memory overrun in high availability configuration with a high number of subscribers
•
Under rare conditions, the standby SCE platform of a cascaded pair crashed and restarted. This happened in a scenario involving a heavy load of anonymous subscribers in cascade topology.
This issue is fixed in SCOS 3.1.6.
Invalid packet generated in MPLS traffic engineering mode with DSCP marking
•
When working in ' MPLS Traffic-Engineering skip ' mode and using DSCP marking, a malformed packet was generated by the SCE when MPLS flows were redirected or blocked.
Flows were affected only if they combined all of the following:
–
–
–
•
DSCP marking+inject+tunneled traffic generated malformed injected packet
•
When working in tos-marking mode and the application injected over tunneled traffic (VLAN/MPLS), the injected packets were malformed since the ToS was updated at the wrong offset in the packet.
This issue is fixed in SCOS 3.1.6.
`show snmp MIB pcube-SE-MIB port' returned wrong number of ports
•
The CLI command ` show snmp MIB pcube-SE-MIB port ' returned the wrong number of ports, because Mng port 2 was treated as traffic port instead of 2nd management port.
This issue is fixed in SCOS 3.1.6.
Applying a policy failed when the SCE was in memory congestion
•
Installing a protocol pack or applying a policy failed when performed under memory congestion and the policy contained many LUT entries (such as URL flavors). An error was issued to the SCA BB Console.
This issue is fixed in SCOS 3.1.6.
Flow filter counters in 'show linecard counters' command were not meaningful to user
•
Output of the ADMIN level command show interface LineCard 0 counters was changed to be more meaningful to the user by specifying traffic-counter names instead of FF counter numbers. The output of this command now shows all traffic counters defined by the user with their names and counting method.
The FF counters previously specified in the output were moved to the ROOT level CLI command show interface LineCard 0 counters flow-filter .
The updated command outputs appear in SCOS 3.1.6.
Simple CLI command needed for globally setting asymmetric Layer 2 support
•
There was no way to globally enable asymmetric layer 2 support. The new CLI command [no] asymmetric-L2-support can now be used to globally enable or disable asymmetric layer 2 support.
Information regarding L2 symmetry was also added to the show flow-open-mode command output.
The new CLI command is supported in SCOS 3.1.6.
FTP password was visible with show version and package extract
•
The FTP password appeared in the output of the show version command and also in the user logs of FTP operations.
This issue is fixed in SCOS 3.1.6.
SCOS RELEASE 3.1.5
Compatibility Information
SCOS 3.1.5 may be installed on the following Service Control Engine platforms:
•
•
•
SCOS 3.1.5 is not compatible with the following Service Control Engine platform:
•
Functional Enhancements
The following section lists the functional enhancements in SCOS release 3.1.5. See the either the Cisco Service Control Engine Software Configuration Guide or the Cisco Service Control Application for Broadband User Guide for more information regarding these features.
•
•
DSCP Marking Enhancements
SCOS release 3.1.5 decouples DSCP marking from the SCE platform queuing mechanism and provides a simplified GUI configuration for DSCP marking based on seven possible DSCP values. After an application is classified by the SCE, the DSCP-marking functionality can mark the relevant packets per package, service, and traffic direction.
Managing MPLS-VPN Branches
SCOS release 3.1.5 extends the functionality of managing an MPLS-VPN encapsulation as a managed subscriber by supporting the ability to define a branch or site as the managed subscriber. The solution provides DPI usage analysis and control per branch of an enterprise in an MPLS-VPN encapsulation.
Identifying Subscribers per VLAN ID and IP
SCOS release 3.1.5 adds the ability to define a subscriber through a combination of VLAN ID and IP address range (subnet).
Flexible Subscriber Introduction
With the introduction of SCA BB release 3.1.5, the Radius Listener LEG component of the Subscriber Manager infrastructure can now leverage a Regular Expressions infrastructure for extracting and manipulating VSA attributes.
Backward Compatibility
MPLS L3-VPN
The 3.1.5 configuration supports a number of different use cases. For more information, please consult the Cisco Service Control Engine (SCE) Software Configuration Guide and the Cisco Service Control MPLS/VPN Solution Guide .
DSCP Marking
The concept of mapping traffic portions to a specific DSCP value has changed in release 3.1.5. In previous versions of the SCA-BB solution, the mapping was only possible based on the CoS (Diffserv Class of Service) to which the Service was mapped. Starting with 3.1.5, it is possible to map each service to one of seven configurable DSCP values independently. The old DSCP marking mode is no longer supported.
Resolved Issues
The following issues were resolved in this release.
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Upgrading the SM caused SCE failover in cascade topology
•
SCE platform performed failover procedure between active and standby SCE platforms in a cascaded pair when the connection to the SM went down, even if no failover behavior was configured for this situation.
This issue is fixed in SCOS 3.1.5.
Non-first fragments misclassified
•
In very rare cases, non-first fragments were misclassified to a different flow and thus potentially dropped.
This issue is fixed in SCOS 3.1.5.
`debug' command caused reboot with fatal error
•
` debug reset-to-default ' command caused a reboot with a fatal error.
This issue is fixed in SCOS 3.1.5.
ip ftp username and password did not work
•
CSCsk06749 The' ip ftp username ' and ` ip ftp password ' commands did not work properly in SCOS 3.1.0
This issue is fixed in SCOS 3.1.5.
Upgrading to PP#11 caused error
•
Upgrading to PP11 both in 3.1.0 and 3.1.1 resulted in the following error message during the first hours after the upgrade:
"Basic Host Context counter below zero!".
This issue is fixed in SCOS 3.1.5.
Defining traffic rules not permitted at the admin authorization level
•
When the ` traffic-rule ' command was executed from the admin authorization level, no traffic rule was created and the following error message appeared:
Error - Required privilege level higher than current level.
This issue is fixed in SCOS 3.1.5.
SCE platform should block simultaneous `get support-file' commands from multiple interfaces
•
If a "logger get support-file" CLI command and a support file extraction operation via the SCA-BB console were executed simultaneously, both operations failed on timeout. Any subsequent support file extraction attempts failed as well.
This issue is fixed in SCOS 3.1.5.
TCP Learning did not close flows that started with SYN-ACK
•
In MPLS/VPN auto-learn mode, upstream SYN-ACK packets of unlearned labels reached the software even though they should not have
This issue is fixed in SCOS 3.1.5.
Could not add VLAN mapping to VPN after removing mapping when subscriber mappings are logged in (3.1.5LA)
•
When a VLAN mapping was removed from a VPN while subscriber mappings (IP@VPN) on the VPN were logged in, subsequent attempts to add a VLAN mapping to that VPN failed with an internal error.
This issue is fixed in SCOS 3.1.5.
Misclassification in MPLS/VPN auto-learn mode
•
In MPLS/VPN auto-learn mode, misclassification and a high ratio of unidirectional flows occurred under certain traffic conditions. Warnings in the debug log from flow tunnel support and/or upstream labels aging were indicative of the problem.
This issue is fixed in SCOS 3.1.5.
Using TIR in MPLS environment caused SCOS - CAM crash
•
Configuring a TIR with IP 0 caused the mechanism that classifies subscribers to TPs to crash.
This issue is fixed in SCOS 3.1.5.
First subscriber had no expiration time (3.1.5LA)
•
The first subscriber logged in via SCE Lease Query LEG had no expiration time.
This issue is fixed in SCOS 3.1.5.
RDRV1 destination startup configuration failed in downgrading from Release 3.1.0 to a previous release
•
When configuring RDR formatter RdrV1 destinations (even when using the old, backward-compatible command syntax), the running-config and hence the startup-config were updated with the new (as of 3.1.0) format of the command, which implicitly states the protocol and transport type. Due to this behavior, in a downgrade scenario, the startup configuration failed to configure the RDR formatter destination, since protocol and transport type are not supported. This resulted in loosing reports.
This issue is fixed in SCOS 3.1.5.
SCE-Sniffer RADIUS LEG did not work after PQI installation
•
After installing a SCA BB PQI file on the SCE platform and before applying a service configuration for the first time, the SCE application ignored all open flows. When a service configuration was applied for the first time, the SCE application started processing new flows. However, older flows that were opened earlier were not processed, and no RDRs were generated for them. RADIUS sniffing is susceptible to this limitation because it is likely that the relevant RADIUS flow would be open before the first time a service configuration is applied
This issue is fixed in SCOS 3.1.5.
Subscribers data synchronization was slower when static subscribers were configured
•
After the cascade links are up, the active box synchronizes the subscriber database to the standby box at a rate of ~1000 updates/sec. However the standby box can support this login rate only for dynamic subscribers. If there are static subscribers configured, it took up to an hour to synchronize the subscriber database after the cascade links were up. However, once the standby box was synchronized, new subscribers were replicated to the standby box normally (at a maximal delay of 2 minutes).
This issue is fixed in SCOS 3.1.5.
CLI command `IP address' did not indicate to the user that a reload of the SCE is required
•
After a change was made to the management IP/subnet, the SCE platform CLI did not output a message notifying the customer that a reload is required for the changes to take effect.
This issue is fixed in SCOS 3.1.5.
SCOS RELEASE 3.1.1
Compatibility Information
SCOS 3.1.1 may be installed on the following Service Control Engine platforms:
•
•
•
SCOS 3.1.1 is not compatible with the following Service Control Engine platform:
•
Resolved Issues
The following issues were resolved in this release.
•
•
•
•
•
RADIUS/DHCP sniffer in SCE might stop functioning for certain flows
•
The SCE stops provisioning RADIUS transactions on certain flows. This can occur in the following situations:
–
–
•
SCE might reload in rare cases due to an infinite loop in its scheduler
•
A bug in the quota export function may cause a subtask to run forever, which may cause the SCE platform to reload.
This issue is fixed in SCOS 3.1.1.
Problem with fragments classification consumes log assets
•
In rare cases a log message of the following type may appear in the log:
[0x0807:0x006c] FC: Total number of packets equals total number of non first fragments in flow countersThis can cause the following issues:
–
–
•
Global-Controllers triggers warnings to the log when querying the MIB Group
•
When querying the global-controllers MIB group counters, the following warning is printed to the log for any global-controller index above 64 and for every port:
<WARNING>[0x0b80:0x000e] SeMib: globalControllersEntry_lookup droppedCounters.actualNumberOfCounters=64 <= gcIndex=64 (port=1)!. <WARNING>[0x0b80:0x000e] SeMib: globalControllersEntry_lookup droppedCounters.actualNumberOfCounters=64 <= gcIndex=65 (port=1)!. ..... <WARNING>[0x0b80:0x000e] SeMib: globalControllersEntry_lookup droppedCounters.actualNumberOfCounters=64 <= gcIndex=1023 (port=1)!.This issue is fixed in SCOS 3.1.1.
Bad response to a SNMP query and a Warning in DBG log
•
When performing a get operation on port number 2 (a management port in rev G), the following message is added to the debug log:
05/04/07 22:55:47 [000000125119:265:475] | 000 | 0000014556 | 0000000 | <WARNING>[0x0b80:0x0011] SeMib: globalControllersEntry_lookup CARDS_GetLimiterBandwidthCfg failed. Error is: interface number must be between 1 and 4.This warning is generated each time the SNMP query is accepted and it means that the SCE responds to the SNMP query with a gen_error code, instead of either responding with a proper value, or responding with a no_such_instance error.
This issue is fixed in SCOS 3.1.1.
SCOS RELEASE 3.1.0
New Features
•
•
Uni-directional classification for support of asymmetric routing
In a situation in which the routing scheme directs the two directions of a flow to follow different routes, each direction flows through a different SCE platform. The effect of different routing is that each SCE platform can classify only one direction of the flow. The `asymmetric routing' mode enables the SCE platform to handle such traffic, allowing SCA BB to classify traffic based on a single direction and to apply basic reporting and global control features to uni-directional traffic.
See "Asymmetric Routing Topology" in Chapter 7, "Configuring the Connection" , in the Cisco Service Control Engine Software Configuration Guide , and the relevant topics in the Cisco Service Control Application for Broadband User Guide
NetFlow V9
Starting from Release 3.1.0 of SCOS for Cisco Service Control, the product can deliver gathered reporting data over the NetFlow V9 export protocol. This protocol is an industry standard for delivering gathered reporting data for external application for collecting, aggregation, storage, and processing. The NetFlow export protocol enables the Service Control solution to integrate with a wide range of existing data collectors and reporters.
See Chapter 8, "Raw Data Formatting: The RDR Formatter and NetFlow Exporting" in the Cisco Service Control Engine Software Configuration Guide .
Subscriber synchronization in cascade setups
SCOS 3.1.0 enhances support of failover in cascade topologies by adding a synchronization process between the active and the standby SCE platforms. This process keeps the standby SCE platform constantly updated with the latest subscriber-related information (login, logout, and quota updates), to minimize information loss if fail-over.
See the section "Synchronizing Subscriber Information in a Cascade System" in Chapter 9, "Managing Subscribers" in the Cisco Service Control Engine Software Configuration Guide .
Compatibility Information
SCOS 3.1.0 may be installed on the following Service Control Engine platforms:
•
•
•
SCOS 3.1.0 is not compatible with the following Service Control Engine platform:
•
Resolved Issues
The following issues were resolved in this release.
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
SSH activity sometimes ended in an unstable state
•
During SSH activity, which usually includes session logout and login operations, the system sometimes ended in an unstable state due to file system violation.
This issue is fixed in SCOS 3.1.0.
Redirect packet was sometimes transmitted on wrong link
•
When a flow using different links for upstream and downstream was redirected, the redirect packet to the subscriber was transmitted on the downstream link, but included the source MAC address from the GET packet seen on the upstream link as destination MAC address. This MAC address was sometimes unknown to the device receiving it.
This issue is fixed in SCOS 3.1.0.
Dynamic Signature (DSS) caused a reboot after dozens of consecutive Apply operations
•
Applying a policy using DSS sometimes caused an infinite loop, resulting in a reboot of the SCE platform.
This issue is fixed in SCOS 3.1.0.
Reload of the SCE due to an expiry of an internal watchdog mechanism
•
Under the following conditions, the SCE platform sometimes reloaded due to an expiry of an internal watchdog mechanism:
–
AND
–
•
The saving process of application configuration was not resilient to reload
•
Running configuration of the application was not saved in cases of sudden reboots. In such cases, the SCE platform came up with no application configuration.
This issue is fixed in SCOS 3.1.0.
Failure to retrieve support file using FTP
•
An error occurred when attempting to retrieve a support file using FTP.
The failure to retrieve the support file was caused by a timeout on the FTP server. When the SCE produces the support file (in zip format), it first produces the contents locally and then adds the contents to the zip file. When the support file is created over an FTP connection, there are long periods of no data transfer during the creation of the zip internal data. These long periods of no data transfer can trigger a connection timeout on the FTP server.
This issue is fixed in SCOS 3.1.0.
PRPC authentication security level was set to default after RPC adapter restart
•
The PRPC connection to the SCE failed after restarting the PRPC server.
This occurred when the security level was changed to something other than the default (semi) and the PRPC server was restarted. After the restart, the security level reverts back to the default value.
This issue is fixed in SCOS 3.1.0.
Part of the quota information was not exchanged between two cascaded SCE platforms
•
Part of the quota information was not exchanged between two cascaded SCE platforms. This caused the failover in SCE cascade topology to be stateless with regard to quota.
This issue is fixed in SCOS 3.1.0 (see Subscriber synchronization in cascade setups).
Subscriber with many mappings did not send all lease time expiration notifications
•
A subscriber with many mappings did not send lease time expiration notification on some mappings.
This issue is fixed in SCOS 3.1.0
Link reflection failed to operate after long uptime
•
Link failure reflection sometimes failed to operate on a system with uptime of more than 24 days.
This issue is fixed in SCOS 3.1.0.
Attack-detector port-list showed random numbers in running-config
•
When a port list was configured for an attack-detector, the specified ports were not saved correctly to the running-config. A list of random numbers was saved.
This issue is fixed in SCOS 3.1.0.
Potential discard of packet under extreme network conditions
•
Under certain conditions, the Flow Filter rule used for congestion handling took effect a few mSec too late, allowing some packets to be discarded.
This issue is fixed in SCOS 3.1.0.
Access violation when configuring anonymous groups
•
The following two scenarios sometimes caused an access violation:
–
Then configure: subscriber anonymous-group name sub1 IP-range 0.0.0.0/0 and transmit traffic.
–
Then remove sub1.
•
destConnectionStatus of the rdr-formatter group was missing in `show snmp MIB pcube-SE-MIB rdr-formatter' command
•
The MIB object destConnectionStatus of the rdr-formatter group was missing in CLI ` show snmp MIB pcube-SE-MIB rdr-formatter ' command.
This issue is fixed in SCOS 3.1.0.
rdrActiveConnectionTrap was not sent after reload of the SCE platform
•
The proprietary trap ` rdrActiveConnectionTrap ', which should be sent after completing a successful establishment of a connection with the RDR collector, was not sent upon system initialization.
This issue is fixed in SCOS 3.1.0.
Link up/down traps were not sent on all ports
•
The SNMP traps 'link up'/'link down', which should be sent for each port upon system initialization, were sent only on the first four ports of the SCE platform.
This issue is fixed in SCOS 3.1.0.
MIB object 'pmoduleType' returned wrong value in cascade setups
•
The MIB object pmoduleType of PcubeSeMib returned the wrong value in cascade setups, indicating that there were only two GBE ports in the SCE platform.
This issue is fixed in SCOS 3.1.0.
MIB variable ifMtu returned wrong value for traffic ports
•
The MIB-II variable ifMtu returned an incorrect value for the maximum packet size of the SCE platform traffic ports.
This issue is fixed in SCOS 3.1.0.
Some rdr-formatter MIB objects returned information only for category 1
•
Global MIB objects in the RdrFormatterGrp in the PcubeSeMib should indicate total values for all categories of the RDR formatter. Previously these objects contained only the values for Category 1.
This issue is fixed in SCOS 3.1.0.
Limitations and Restrictions
The upgrade to the SCOS 3.1.6 release may result in re-initialization of the SCE 1010 or SCE 2020 hardware Bypass module. This re-initialization process may cause a failure of the GBE link where the system stalls for a period of less than 1 sec.
The table below states the various cases when this re-initialization may occur (marked as "Yes").
•
The following is recommended:
•
•
Use the following root-level CLI commands to disable the SCE watchdog:
configure watchdog software-reset disabled interface linecard 0 no watchdog•
configure watchdog software-reset enabled interface linecard 0 watchdogOpen Caveats
•
•
•
•
•
•
•
•
•
•
•
•
•
SCE reloaded due to watchdog expiry
•
The SCE is sometimes forced to reload due to a watchdog expiry. This occurs under the following conditions:
–
–
An error dump to the log that takes more then 1mSec triggers the traverser watchdog. It seems that the error handling of the traverser watchdog during the execution of a different error handling causes an invalid memory access, which results in a timeout of the system watchdog. It is the timeout of this system watchdog that causes the SCE platform reload.
The traverser watchdog is a mechnanism that protects the SCE from infinite application loops in the context of a single flow and atttempts to kill that single flow rather than reload the SCE (which is what occurs when the traverser watchdog is disabled). There are cases in which, for debug purposes, disabling of this mechanism or extension of its configured timeout is desireable.
Note
Workaround: Extend the timeout of the traverser watchdog by editing the DbSched.txt file, which is located under /tffs0/system/p3hidden/.
1.
2.
lcConstDb.traverserWatchdog.timeOutMicroSec 50003.
4.
5.
SCE#>debug db show-all lcconstdb.traverserwatchdog.time # IC Name Size (content of first element) 1322 C lcConstDb.traverserWatchdog.timeOutMicroSec 1 (= '5000'')Refinement of the maximum number of IP mappings per subscribers
•
Subscriber data sync between cascaded SCE platforms may potentially cause a message overflow, if there are a large number of IP mappings per subscriber. When more than 200 IP mappings are found for a given subscriber, the resulting message overflow might corrupt the stack.
Workaround: Make sure no subscriber is introduced with more than 200 IP mappings.
Unused ACLs disappear after applying policy
•
An ACL that is configured but not applied to anything (such as an interface, access-map, or snmp community string) might disappear from the configuration after applying a policy to the SCE platform.
Workaround: Disable the Anomaly Detection capability (part of the Security dashboard).
When L2TP skip is configured, non L2TP fragments may cause flow mismatch
•
Non-first fragments of pure IP traffic (not tunneled) are not handled correctly when the system is in L2TP skip mode. Incorrect UDP/TCP ports are assumed, and the fragment is mapped to the wrong flow.
Workaround: configure traffic rules that will correctly define the ports. Refer to the CLI sample for specific command formats.
1.
2.
3.
4.
5.
configure interface LineCard 0 traffic-counter name tcp_frag count-packets traffic-counter name udp_frag count-packets traffic-rule name quick-forward-all IP-addresses all protocol all direction both traffic-counter none action quick-forwarding traffic-rule name tcp_frag IP-addresses all protocol TCP ports subscriber-side 0 network-side 0 flags all direction both traffic-counter name tcp_frag action ignore traffic-rule name udp_frag IP-addresses all protocol UDP ports subscriber-side 0 network-side 0 direction both traffic-counter name udp_frag action ignore exit exit copy running-config-all startup-config-all reloadQuota events not received by SCE subscriber API client or QM
•
Quota events are not received by the SCE subscriber API client or QM because the internal RDR connection to destination 127.0.0.1 port 33001 is not configured.
Workaround: configure the internal RDR connection as follows:
1.
2.
While removing all VPNs management operations cannot be performed
•
When removing all VPNs from the SM using the " --force " option, some management operations cannot be performed on the SCE until the operation completes. This occurs only when removing many VPNs that have active subscriber mappings in the SCE.
Workaround: Instead of removing the VPNs along with their subscriber mappings by using the --force option, remove the subscribers first, and only then remove the VPNs (without the --force option.
Tunnel-id-based traffic rule applies DSCP marking to non-tunneled traffic
•
A tunnel-id-based traffic rule defining DSCP marking applies the DSCP marking to non-tunneled traffic, also. Workaround:
Workaround: When defining the traffic rule, always set the URG flag. For existing rules, replace with a new rule that is identical, with the addition of setting the URG flag.
SCE2000 may reload if stalled by flow control
•
By design, the SCE platform reacts to Ethernet flow control and doesn't activate it. Therefore, it is possible for a situation to arise in which flow control stalls the SCE platform by overflowing the SCE platform queues and thereby causing traffic to be dropped on the Rx interfaces. If this situation persists for more than five seconds, it may trigger the SCE platform internal sanity checks mechanism, which may in turn trigger a reload of the SCE platform in an attempt to recover.
SCE2000 may drop cascade control packets from AF4 queue during GC congestion
•
Extra care must be given to the configuration of the link shapers in 'inline-cascade' connection mode. Configuring the shaper in an aggressive manner might result in very high rate of tail-dropped packets. In extreme situations, packets that are used for the High Availability protocol monitoring and control may be dropped. Thus, an extreme situation could result in false detection of a failure and an unnecessary switchover between the active and standby SCE platforms
Link failure may be reflected to all ports if a port is flickering due to a HW problem
•
When link reflection on all ports with linecard aware is configured, the link failure may be reflected to all ports (rather than only to the relevant link) if one of the ports that is connected to the failed linecard is flickering due to a hardware problem.
The configured attack threshold is set for each PPC separately
•
For certain types of attacks, an attack is detected by the SCOS attack-filter module only if it is three times stronger (as measured by flow rate per second) than the configured value.
This happens when the IP address common to all the flows of the attack is on the network side of the SCE platform, so all attacks of type 'single-side-network' have this problem
When the VAS Health Check initializes, the CLI command `show interface linecard 0 VAS-traffic-forwarding VAS server-id <id>' shows the server being UP even if it is actually Down
•
The operative state of a VAS server while the Health Check is in Init state is considered to be Up as shown in the CLI command " show interface linecard 0 VAS-traffic-forwarding VAS server-id <id>". In addition, during this time, the SCE platform may forward VAS traffic to this server.
Flow `opened from VAS' is misrouted if there is a FF rule to bypass
•
When VAS mode is enabled, the system generally assumes that traffic with a VLAN tag is VAS traffic coming from the VAS servers, and therefore forwards it to the non-VAS link.
However, under the following conditions, a flow will be forwarded by the SCE platform on the same link on which it was received and with no VLAN tag:
–
–
–
•
Packet Loss during Application Installation or Upgrade
•
When a PQI application file is installed or upgraded on the SCE, the SCE may lose a few packets for a few seconds. The overall percentage of this phenomenon is very low.
Workaround : It is advised to perform the upgrade in non peak time.
Obtaining Documentation and Submitting a Service Request
For information on obtaining documentation, submitting a service request, and gathering additional information, see the monthly What's New in Cisco Product Documentation, which also lists all new and revised Cisco technical documentation, at:
http://www.cisco.com/en/US/docs/general/whatsnew/whatsnew.html
Subscribe to the What's New in Cisco Product Documentation as a Really Simple Syndication (RSS) feed and set content to be delivered directly to your desktop using a reader application. The RSS feeds are a free service and Cisco currently supports RSS version 2.0.
CCDE, CCENT, CCSI, Cisco Eos, Cisco HealthPresence, Cisco Ironport, the Cisco logo, Cisco Lumin, Cisco Nexus, Cisco Nurse Connect, Cisco Stackpower, Cisco StadiumVision, Cisco TelePresence, Cisco Unified Computing System, Cisco WebEx, DCE, Flip Channels, Flip for Good, Flip Mino, Flip Video, Flip Video (Design), Flipshare (Design), Flip Ultra, and Welcome to the Human Network are trademarks; Changing the Way We Work, Live, Play, and Learn, Cisco Store, and Flip Gift Card are service marks; and Access Registrar, Aironet, AsyncOS, Bringing the Meeting To You, Catalyst, CCDA, CCDP, CCIE, CCIP, CCNA, CCNP, CCSP, CCVP, Cisco, the Cisco Certified Internetwork Expert logo, Cisco IOS, Cisco Press, Cisco Systems, Cisco Systems Capital, the Cisco Systems logo, Cisco Unity, Collaboration Without Limitation, EtherFast, EtherSwitch, Event Center, Fast Step, Follow Me Browsing, FormShare, GigaDrive, HomeLink, Internet Quotient, IOS, iPhone, iQuick Study, IronPort, the IronPort logo, LightStream, Linksys, MediaTone, MeetingPlace, MeetingPlace Chime Sound, MGX, Networkers, Networking Academy, Network Registrar, PCNow, PIX, PowerPanels, ProConnect, ScriptShare, SenderBase, SMARTnet, Spectrum Expert, StackWise, The Fastest Way to Increase Your Internet Quotient, TransPath, WebEx, and the WebEx logo are registered trademarks of Cisco Systems, Inc. and/or its affiliates in the United States and certain other countries.
All other trademarks mentioned in this document or website are the property of their respective owners. The use of the word partner does not imply a partnership relationship between Cisco and any other company. (0907R)
Any Internet Protocol (IP) addresses used in this document are not intended to be actual addresses. Any examples, command display output, and figures included in the document are shown for illustrative purposes only. Any use of actual IP addresses in illustrative content is unintentional and coincidental.
© 2009 Cisco Systems, Inc. All rights reserved.
