By Leland Chen-Oracle on Aug 20, 2014
The cluster event SNMP interface was first introduced in Oracle Solaris Cluster 3.2 release. The details of the SNMP interface are described in the Oracle Solaris Cluster System Administration Guide and the Cluster 3.2 SNMP blog.
Prior to the Oracle Solaris Cluster 4.2 release, when the event SNMP interface was enabled, it would take effect on WARNING or higher severity events. The events with WARNING or higher severity are usually for the status change of a cluster component from ONLINE to OFFLINE. The interface worked like an alert/alarm interface when some components in the cluster were out of service (changed to OFFLINE). The consumers of this interface could not get notification for all status changes and configuration changes in the cluster.
Cluster Event and its SNMP Interface in Oracle Solaris Cluster 4.2
The user model of the cluster event SNMP interface is the same as what was provided in the previous releases. The cluster event SNMP interface is not enabled by default on a freshly installed cluster; you can enable it by using the cluster event SNMP administration commands on any cluster nodes. Usually, you only need to enable it on one of the cluster nodes or a subset of the cluster nodes because all cluster nodes get the same cluster events. When it is enabled, it is responsible for two basic tasks.
• Logs up to 100 most recent NOTICE or higher severity events to the MIB.
• Sends SNMP traps to the hosts that are configured to receive the above events.
The changes in the Oracle Solaris Cluster 4.2 release are
1) Introduction of the NOTICE severity for the cluster configuration and status change events.
The NOTICE severity is introduced for the cluster event in the 4.2 release. It is the severity between the INFO and WARNING severity. Now all severities for the cluster events are (from low to high)
• INFO (not exposed to the SNMP interface)
• NOTICE (newly introduced in the 4.2 release)
In the 4.2 release, the cluster event system is enhanced to make sure at least one event with the NOTICE or a higher severity will be generated when there is a configuration or status change from a cluster component instance. In other words, the cluster events from a cluster with the NOTICE or higher severities will cover all status and configuration changes in the cluster (include all component instances). The cluster component instance here refers to an instance of the following cluster components
node, quorum, resource group, resource, network interface, device group, disk, zone cluster and geo cluster heartbeat. For example, pnode1 is an instance of the cluster node component, and oracleRG is an instance of the cluster resource group.
With the introduction of the NOTICE severity event, when the cluster event SNMP interface is enabled, the consumers of the SNMP interface will get notification for all status and configuration changes in the cluster. A thrid-party system management platform with the cluster SNMP interface integration can generate alarms and clear alarms programmatically, because it can get notifications for the status change from ONLINE to OFFLINE and also from OFFLINE to ONLINE.
2) Customization for the cluster event SNMP interface
• The number of events logged to the MIB is 100. When the number of events stored in the MIB reaches 100 and a new qualified event arrives, the oldest event will be removed before storing the new event to the MIB (FIFO, first in, first out). The 100 is the default and minimum value for the number of events stored in the MIB. It can be changed by setting the log_number property value using the clsnmpmib command. The maximum number that can be set for the property is 500.
• The cluster event SNMP interface takes effect on the NOTICE or high severity events. The NOTICE severity is also the default and lowest event severity for the SNMP interface. The SNMP interface can be configured to take effect on other higher severity events, such as WARNING or higher severity events by setting the min_severity property to the WARNING. When the min_severity property is set to the WARNING, the cluster event SNMP interface would behave the same as the previous releases (prior to the 4.2 release).
• Set the number of events stored in the MIB to 200
# clsnmpmib set -p log_number=200 event
• Set the interface to take effect on WARNING or higher severity events.
# clsnmpmib set -p min_severity=WARNING event
Administering the Cluster Event SNMP Interface
Oracle Solaris Cluster provides the following three commands to administer the SNMP interface.
• clsnmpmib: administer the SNMP interface, and the MIB configuration.
• clsnmphost: administer hosts for the SNMP traps
• clsnmpuser: administer SNMP users (specific for SNMP v3 protocol)
Only clsnmpmib is changed in the 4.2 release to support the aforementioned customization of the SNMP interface. Here are some simple examples using the commands.
1. Enable the cluster event SNMP interface on the local node
# clsnmpmib enable event
2. Display the status of the cluster event SNMP interface on the local node
# clsnmpmib show -v
3. Configure my_host to receive the cluster event SNMP traps.
# clsnmphost add my_host
Cluster Event SNMP Interface uses the common agent container SNMP adaptor, which is based on the JDMK SNMP implementation as its SNMP agent infrastructure. By default, the port number for the SNMP MIB is 11161, and the port number for the SNMP traps is 11162. The port numbers can be changed by using the cacaoadm. For example,
# cacaoadm list-params
Print all changeable parameters. The output includes the snmp-adaptor-port and snmp-adaptor-trap-port properties.
# cacaoadm set-param snmp-adaptor-port=1161
Set the SNMP MIB port number to 1161.
# cacaoadm set-param snmp-adaptor-trap-port=1162
Set the SNMP trap port number to 1162.
The cluster event SNMP MIB is defined in sun-cluster-event-mib.mib, which is located in the /usr/cluster/lib/mibdirectory. Its OID is 126.96.36.199.188.8.131.52.80, that can be used to walk through the MIB data. Again, for more detail information about the cluster event SNMP interface, please see the Oracle Solaris Cluster 4.2 System Administration Guide.
- Leland Chen