Cisco MCS-7825-H3-IPC1 Service Manual page 153

Chapter 4 Simple Network Management Protocol
–
–
Because System Application Agent cannot show services that are activated and deactivated or monitor
Web App services or servlets, use this approach to monitor system health and service status for Cisco
Unified Communications Manager applications:
• Use the Serviceability API getservicestatus to provide complete status information, including
activation status, for both Web applications and non-Web applications. See the AXL Serviceability
API Guide for more details.
Check service status with this CLI command: utils service list
•
Monitor the servM-generated messages with Syslog (see the following example):
•
Mar 18 16:40:52 ciscart26 local7 6 : 92: Mar 18 11:10:52.630 UTC :
%CCM_SERVICEMANAGER-SERVICEMANAGER-6-ServiceActivated: Service Activated. Service
Name:Cisco CallManager SNMP Service App ID:Cisco Service Manager Cluster ID: Node
ID:ciscart26
If an SNMP request specifies multiple OIDs and the variables are pointing to empty tables, you may get
a NO_SUCH_NAME (for SNMP V1) or GENERIC ERROR (for SNMP V2c or V3) due to a timeout
problem. A timeout can occur as a result of throttling enhancements to protect the Cisco Unified
Communications Manager processing engine.
You can retrieve the count of entries in CCMH323DeviceTable and ccmSIPDeviceTable by using scalar
objects, so the SNMP Manager (the client) can avoid unnecessary get/getnext operations on these tables
when no entries exist. As an SNMP developer, you can use the following workaround for this problem:
Use the available scalar variables (1.3.6.1.4.1.9.9.156.1.5) to determine table size before accessing
•
the table or perform the get operation on the desired table; then, query the non-empty tables.
Reduce the number of variables that are queried in a single request; for example, for empty tables,
•
if the management application has the timeout set to 3 seconds, specify only 1 OID. (For non-empty
tables, it takes 1 second to retrieve one row of data.)
Increase the response timeout.
•
Reduce the number of retries.
•
Avoid using getbulk SNMP API. The getbulk API retrieves the number of records that is specified
•
by MaxRepetitions, so even if the next object goes outside the table or MIB, it gets those objects.
Empty tables cause even more delay. Use getbulk API for non-empty tables with a known number
of records. In these circumstances, set MaxRepetitions to 5 seconds to require a response within 5
seconds.
Structure SNMP queries to adapt to existing limits.
•
Avoid performing multiple getbulks to walk the PhoneTable periodically in case a large number of
•
phones are registered to Cisco CallManager. You can use the ccmPhoneStatusUpdateTable, which
updates whenever there is a Phone update, to decide whether to walk the PhoneTable.
OL-22523-01
ccmPhoneFailedAlarmInterval (1.3.6.1.4.1. 9.9.156.1.9.2) set to 30-3600. You can use this CLI
command: snmpset -c <community string> -v2c <transmitter ipaddress>
1.3.6.1.4.1.9.9.156.1.9.2 .0 i <value>
ccmPhoneStatusUpdateAlarmInterval (1.3.6.1.4.1.9.9.156.1.9.4) set to 30-3600. You can use
this CLI command: snmpset -c <community string> -v2c <transmitter ipaddress>
1.3.6.1.4.1.9.9.156.1.9.4.0 i <value>
Verify that you configured the notification destination properly in the Notification Destination
(V1/V2c or V3) Configuration window.
Verify that you configured the community string/user privileges correctly, including Notify
permissions, in the Community String (V1/V2c) or User (V3) Configuration window.
Cisco Unified Communications Manager Managed Services Guide
SNMP Troubleshooting
4-7