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.

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 on page 11 .

You can also monitor the service node counters. See the “Displaying Per Service Node and Service Node Group Statistics” section on page 6 .

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.

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

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.

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

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.

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.

Some older service nodes may not support all application accelerators.

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

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.

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.

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.