Configuration Guide for AppNav-XE on the Cisco CSR 1000V Series and Cisco ASR 1000 Series
Troubleshooting
Downloads: This chapterpdf (PDF - 318.0KB) The complete bookPDF (PDF - 1.21MB) | Feedback

Table of Contents

Troubleshooting

Using Debug Commands

AppNav-XE Debug Commands

Clearing AppNav-XE Statistics

Debugging the Cisco IOS-XE Control Plane

Debugging the Cisco IOS-XE Infrastructure

Debugging the Data Plane

AppNav Service Node Auto Discovery Debug Commands (For CiscoCSR 1000V Series Only)

Common Problems

Traffic Not Redirected

Traffic Passed Through Instead of Redirected

Degraded Cluster

Service Node Excluded

Flows Not Synced Between AppNav Controllers

Connection Hangs

Connection Resets

Application Accelerator Status Shows as Red with No Load

The AppNav-XE Component Fails to Initialize

Flow Limit Reached

Other AppNav-XE Known Issues

Troubleshooting

This chapter explains how to troubleshoot common problems and covers the following sections:

Using Debug Commands

Clearing AppNav-XE Statistics

To clear all the AppNav-XE statistics or just certain statistics, use the following command:

router# clear service-insertion statistics ?
 
all Clear all service-insertion statistics
appnav-controller Clear appnav-controller statistics
appnav-controller-group Clear appnav-controller-group statistics
service-context Clear service-context statistics
service-node Clear service-node statistics
service-node-group Clear service-node-group statistics

Debugging the Cisco IOS-XE Control Plane

Use the following debug commands to trace control plane activities:

router# debug appnav-controller ?
cm-reg Debugging AppNav Controller CM registration with the WCM server
cmm Debugging AppNav Controller Cluster Management (CMM)
fdm Debugging AppNav Controller Flow Distribution Module (FDM)
ha Enable AppNav Controller high availability (HA) redundancy checkpoint and ISSU
infrastructure debugs
vi Debugging AppNav Controller Virtual Interface (VI), including the status at the
time of creation and links to the compress and uncompress interface
 
router# debug appnav-controller cmm ?
all Enable all CMM debugs
cli Enable CMM CLI debugs
events Enable CMM state machine event debugs
misc Enable CMM misc debugs
packets Reception and transmission of packets (can be filtered based on IP address)
shell Enable CMM misc debugs
timers Enable CMM misc debugs
 
router# debug appnav-controller fdm ?
all Enable all FDM debugs
events Enable debugging for important events being handled by FDM
infra Enable debugging for FDM infrastructure events
 

The following debug commands are the most useful:

  • debug appnav-controller cmm events
  • debug appnav-controller fdm events
  • debug appnav-controller ha

Showing Packet Drop Statistics

Use the following command to display unplanned packet drops:

router# show platform hardware qfp active statistics drop
-------------------------------------------------------------------------
Global Drop Stats Packets Octets
-------------------------------------------------------------------------
AppNavBadRoute 38 2888
Ipv4AclLookupMiss 42 3034
Ipv4NoRoute 4408 1293334
UnconfiguredIpv4Fia 19 1710
 

The following are the reasons for which packets may drop:

  • AppNavInvSNpkt—Malformed or unsupported packet from the service node.
  • AppNavInternalErr—Logic error within the AppNav-XE component. Uncommon.
  • AppNavBadRoute—A non-AppNav-XE packet appeared at the AppNav-XE virtual or tunnel interface. Very common when routing protocols are enabled.
  • AppNavNoTunnel—There is no tunnel facility available for the service node-bound packet.
  • AppNavNoSvcCtx—There is no service context matching the flows from the service node.
  • AppNavInvFOState—The flow state is no longer valid. This is usually due to changes in the configuration.
  • AppNavUnexpctdpkt—The AppNav-XE component did not expect to process more packets because it has been shut down.

Showing Data Path CPU Utilization

To display the data path CPU utilization, use the following command:

router# show platform hardware qfp active datapath utilization
 
CPP 0: Subdev 0 5 secs 1 min 5 min 60 min
Input: Priority (pps) 0 0 0 0
(bps) 0 72 88 48
 
Non-Priority (pps) 226455 225968 198785 72441
(bps) 1879325304 1875408168 1648044616 599951168
 
Total (pps) 226455 225968 198785 72441
(bps) 1879325304 1875408240 1648044704 599951216
 
Output: Priority (pps) 229023 228474 245267 90057
(bps) 1619093520 1641710256 2389617496 949076160
 
Non-Priority (pps) 209522 208053 293300 104501
(bps) 180090080 178161632 3124680344 1191566064
 
Total (pps) 438545 436527 538567 194558
(bps) 1799183600 1819871888 5514297840 2140642224
 
Processing: Load (pct) 26 26 19 8

Showing Data Path Memory Utilization

Use the following command to show statistics about the data path memory use.


Note The value for In Use DRAM memory must be less than 90 percent of the value for Total DRAM memory; otherwise, the AppNav-XE component stops optimizing new flows.


router# show platform hardware qfp active infrastructure exmem statistics
 
QFP exmem statistics
Type: Name: DRAM, QFP: 0
Total: 268435456
InUse: 99933184
Free: 168502272
Lowest free water mark: 168502272
 
Type: Name: IRAM, QFP: 0
Total: 134217728
InUse: 8087552
Free: 126130176
Lowest free water mark: 126130176
 
Type: Name: SRAM, QFP: 0
Total: 32768
InUse: 15088
Free: 17680
Lowest free water mark: 17680

Debugging the Data Plane

The output of the following debug command is displayed as a log file named /tmp/fp/trace/cpp_cp_ Fx -0.log under the FP shell, where Fx is either F0 or F1 depending on the active FP module. You need a shell license to access the FP shell.

If you do not have shell access, you can use the test platform software trace slot fp act cpp-control-process rotate command to force the log to flush to bootflash:tracelogs.

router# debug platform hardware qfp active feature appnav-controller datapath ?
classify Debug QFP flow classification such as traces, policy, peer ID, and
classification action (which service node group)
drop Enable drop debugging and shows traces of packet drop due to errors
fdl Debug QFP flow distribution such as selecting a service node within a service
node group
ha Debug QFP high availability (HA) and AppNav Controller issues. Shows traces
related to syncing flows between AppNav Controllers and between active and
standby FPs
interop Debug QFP feature interoperations such as FNF, NBAR, and NAT
pkt-path Debug QFP packet processing and packet interception
proxy Debug QFP proxy issues related to interface with the control place, such as
statistics reporting and configuration
 

Each of the above categories (other than drop , which has no level) has the following four levels:

  • Error—Displays error level debugs and detects potential issues.
  • Warn—Displays warnings and errors.
  • Info—Displays information, warnings, and errors.
  • All—The lowest level of debugging. Displays all debugs.

To limit the number of debug messages, we recommend that you only enable the error debug level first and then slowly reduce the debug level.

You can also use the following command to check on packets dropped by the router. The command lists all the packets that were dropped with a reason. If you see AppNav drop reasons, you can enable the debug drop command to see the actual packet drops inside the trace logs.

router# show platform hardware qfp active statistics drop
 
Global Drop Stats Packets Octets
------------------------------------------------------
The Global drop stats were all zero

AppNav Service Node Auto Discovery Debug Commands (For Cisco CSR 1000V Series Only)

Use the following debug commands to trace the AppNav service node auto discovery feature on the Cisco CSR 1000V Series:

  • debug appnav auto-discovery
  • debug mdns all
  • debug mdns packet

Note The Auto Discovery debug commands are applicable to the Cisco CSR 1000V Series only.


Common Problems

Traffic Not Redirected

If traffic is not redirected properly, ensure that “service-insertion waas” is present on interfaces on which the traffic is supposed to be intercepted. Issue the show service-insertion status command to verify this.

Traffic Passed Through Instead of Redirected

The show service-insertion statistics connection command indicates whether traffic is passed through or redirected. If traffic is passed through instead of being redirected, use the show policy-map target service-context context_name passthru-reason command to find out the reason. For details, see the “Displaying Pass Through Reason Statistics” section.

You can also monitor the service node counters. See the “Displaying Per Service Node and Service Node Group Statistics” section.

The term “Initial Redirect” indicates that flows are being redirected to the service nodes. If the flows are not being redirected to the service nodes, maybe the policy did not cover the traffic type.

The “Initial Redirect -> Passthrough” counter indicates that the service node has decided to pass-through the flow. This is likely due to policies on the service node.

The “Redirect -> Passthrough” counter indicates that the service node later decided to pass-through the flow. This is likely due to lack of a peer WAAS device. Two WAAS devices are needed along the path to optimize a flow.

Degraded Cluster

If connections are passed through and you are using an AppNav Controller group that has two or more AppNav Controllers, it is possible that the cluster state is degraded instead of operational. This means that the AppNav Controller view is not the same on each of the AppNav Controllers.

To check the cluster state and the stable AppNav Controller view on each of the AppNav Controllers, use the following command:

router# show service-insertion service-context
 
Service Context : waas/1
Cluster protocol ICIMP version : 1.1
Cluster protocol DMP version : 1.1
Time service context was enabled : Fri Dec 7 19:28:11 2012
Current FSM state : Degraded
Time FSM entered current stat : Fri Dec 7 21:58:29 2012
Last FSM state : Converging
Time FSM entered last state : Fri Dec 7 21:58:19 2012
Cluster operational state : Degraded
 
Stable AppNav controller View:
21.0.0.145
21.0.0.160
 
Stable SN View:
21.0.0.149
 

The reason for the difference in AppNav Controller views on the AppNav Controllers may be due to a mismatch in the AppNav Controller group configuration on the AppNav Controllers or due to a connectivity problem between the AppNav Controllers.

It is also useful to check the alarms on each of the AppNav Controllers by using the following command that also suggests corrective actions:

router# show service-insertion alarms detail support
 
Critical Alarms:
----------------
Alarm Instance Alm ID Module AC/SN IP Addr AO SNG
1 degraded_cluster 29002 cmm N/A N/A N/A
 
Cluster protocol detected inconsistency in AC view of peer ACs. Device will pass-through all new connections.
 
Cluster view is degraded.
 

Explanation:

Cluster membership manager has detected a discrepancy in the AC view of peer ACs. Optimization will be turned off on this device for cluster consistency.

Action:

Check the network for partial connectivity issues.
 
Major Alarms:
-------------
Alarm Instance Alm ID Module AC/SN IP Addr AO SNG
1 ac_unreachable 29006 cmm 192.168.1.11 N/A N/A
 
Cluster protocol on device cannot communicate with peer AC ("192.168.1.11").
 

AppNav controller is unreachable.

 

Explanation:

Cluster protocol detected failure of the peer AC. This could happen due to several reasons - configuration mismatch or network issues preventing communication between the ACs or the AC actually being down.

Action:
Other alarms will indicate if this is a configuration issue. If so, correcting the configuration mismatch will cause this alarm to go away. Otherwise, check the network to see if the devices are able to communicate with each other.
 
Minor Alarms:
-------------
None

Service Node Excluded

If no traffic is redirected to a particular service node and you are using an AppNav Controller group with two or more AppNav Controllers, it is possible that the service node is excluded. This happens when the service node view is not the same on each of the AppNav Controllers.

To check the stable service node view on each of the AppNav Controllers, use the show service-insertion service-context command.

The reason for the difference in service node views could be due to a mismatch in the service node group configuration on the AppNav Controllers or due to a connectivity problem between one or more of the AppNav Controllers and the excluded service node.

To check if any service nodes are excluded or unreachable, look for the SN_excluded and SN_unreachable alarms by using the show service-insertion alarms detail support command on each of the AppNav Controllers.

Flows Not Synced Between AppNav Controllers

This could be due to a mismatch in the VRF names for the traffic seen by the AppNav Controllers in the ACG.

Check the output of the show service-insertion statistics connection summary command for the counter for Flow Sync Failures due to vrf mismatch.

router# show service-insertion statistics connection summary
Number of 2T entries=0
Number of 3T entries=0
Number of optimized flows=0
Number of pass-through flows=0
Flow sync failures due to vrf-mismatch=3
 

Connection Hangs

A connection might be considered “hung” for various reasons. In many cases, it helps to use telnet to simulate a connection to the server. For example, enter telnet HTTP_server 80 .

If the connection hangs during the TCP 3-way handshake, verify that both the connection and the route to the service node are properly set up.

If the connection hangs after the connection was established, verify the connection along the path. Make sure that the MTU along the path is correct.

Use the show service-insertion statistics connection command on the AppNav Controller and the show statistics connection command on the service node to cross check the connections between the AppNav Controller and the service node.

Use the show platform hardware qfp active statistics drop command to check for packet drops.

Connection Resets

You can usually see the reason for the connection reset by issuing the show statistics connection closed command and the show statistics connection closed conn-id connection_ID command on the service node. Capturing packets is also useful in analyzing the reason for the connection reset.

Use the show platform hardware qfp active statistics drop command to check for dropped packet.

Application Accelerator Status Shows as Red with No Load

Some older service nodes may not support all application accelerators.

Individual application accelerators, such as the video application accelerator, require a separate license.

The AppNav-XE Component Fails to Initialize

If the system displays an ERROR_NOTIFY syslog message when you enable the service-insertion waas command on the interface, it could be that the AppNav-XE component failed to initialize due to low memory. Check the amount of memory by using the following command:

router# show platform hardware qfp active infrastructure exmem statistics
QFP exmem statistics
 
Type: Name: DRAM, QFP: 0
Total: 268435456
InUse: 102283264
Free: 166152192
Lowest free water mark: 166152192
Type: Name: IRAM, QFP: 0
Total: 134217728
InUse: 8186880
Free: 126030848
Lowest free water mark: 126030848
Type: Name: SRAM, QFP: 0
Total: 32768
InUse: 15088
Free: 17680
Lowest free water mark: 17680
 

If the available memory is less than 10 percent of the total memory, the AppNav-XE component may not be able to initialize, which results in no flows being redirected.

If the output of the show policy-map target service-context waas/1 command is blank, instead of listing the AppNav policy being used, it may indicate that the system was unable to initialize.

Flow Limit Reached

Both the AppNav Controller and the service nodes have a limit on the number of flows that they can support. On the AppNav Controller, the limit is 2 million flows. Beyond that, all flows are passed through. If you exceed the limit, the system displays the following error message:

03/10 00:53:51.720 [errmsg]: (warn): %CFT_CLIENT-4-MAX_FCS_TOUCH_WARN: CFT number of flow-context threshold is reached, can't allocate more memory for flow-context.
 

The flow limit may be reached in advance due to available memory. In this case, the system displays the following syslog message:

*Aug 24 00:29:17.205: %CFT_CLIENT-4-CFT_MEMORY_BOUNDARY_TOUCH_WARN: F0: cpp_cp: CFT reached maximum configured memory utilization. Can't allocate more memory for flow-context.
 

In both cases, when the existing flows are completed and the number of flows dips below the threshold, flows are optimized again.

Other AppNav-XE Known Issues

If the AppNav Controller does not respond to a WAAS TCP trace, the system forwards the TCP trace to the service node and the service node generates a response along with a list of service nodes along the path.