Hardware Management Agent Configuration File

Language:

By default, the Hardware Management Agent manages the storage poller and provides Oracle ILOM information about storage devices in the system including any changes to these devices.

In Oracle Solaris 11.3 SRU 18, a feature was added that allows the trap proxy to notify the agent when Oracle ILOM has sent a trap. Instead of regularly polling Oracle ILOM for information, detection of a trap will initiate an update cycle to get the newest data from Oracle ILOM. By default, if no traps trigger a polling cycle then Oracle ILOM is polled by the Hardware Management Agent once per hour.

Note - The trap-based polling feature will only work if the host is configured for SNMP monitoring and the Oracle LOM trap proxy has been set up. For information on setting up the Oracle ILOM trap proxy, see Configuring Oracle ILOM SNMP Trap Forwarding Using itpconfig.

If more precise real-time data of sensor values (temperature/voltage) is required, a custom "polling_round_delay" configuration option can be added to the /etc/ssm/hwmgmtd.conf file. Adding this option into the configuration file allows you to change the default polling interval.

For SNMP monitoring, the Hardware Management Agent can be configured as an SNMP agent for the SUN-HW-MONITORING-MIB and SUN-STORAGE-MIB that are part of Oracle Hardware Management Pack. This enables the native SNMP service running on the host to query information stored in these MIBs. By default, this feature is turned off but can be configured in the hwmgmtd.conf configuration file.

To prevent possible memory leaks that can cause performance issues, the Hardware Management Agent service (svc:/system/sp/management:default) can be configured to perform a daily reset at a specified time.

The Hardware Management Agent also records log messages in the /var/log/ssm/hwmgmtd.log file. These messages can be used to troubleshoot the running status of the Hardware Management Agent. Log level configuration is done using basic or bit flag parameters.

The following subsections describe the configurable parameters in the hwmgmtd.conf file. For information on performing the configuration, see Configure the Hardware Management Agent for SNMP Support and Logging Level.

Management Agent Daily Reset, Polling and SNMP Support

The following table lists the configurable values for polling and thread selection in the hwmgmtd.conf file when SNMP monitoring is used on the host. In addition, there is a daily reset setting for the management agent to prevent possible memory leak issues.

Note - The default configuration for polling provides the optimal functionality in systems where SNMP monitoring is not used. Where host-based SNMP monitoring is configured and the Oracle ILOM trap proxy setup using itpconfig, the Hardware Management Agent will perform an initial poll of Oracle ILOM and then listen for Oracle ILOM SNMP traps to decide when a re-pole is needed. Enabling SNMP polling threads increase resources used by the Hardware Management Agent which might affect system performance.

Thread	Description
storage_poller_enabled=<value>	When enabled, the Hardware Management Agent provides up-to-date information about host internal storage to Oracle ILOM. In addition, if SNMP monitoring is enabled on the host, this poller responds to host SNMP requests to the SUN-STORAGE-MIB. Supported values are: 0=disabled; 1=enabled. The default is 1. Note - Disabling this parameter is not recommended as it will prevent storage information from being sent to Oracle ILOM.
inventory_poller_enabled=<`value`>	When enabled, this poller responds to host SNMP requests to the SUN-HW-MONITORING-MIB. Supported values are: 0=disabled; 1=enabled. The default is 0.
request_thread_enabled=<`value`>	When enabled, the Hardware Management Agent responds to host SNMP requests and listens for SNMP traps generated by Oracle ILOM. Supported values are: 0=disabled; 1=enabled. The default is 0.
polling_round_delay=<`time in seconds`>	If needed, this option can be added. When set, this option overrides the default polling delay of every hour, or sooner if a trap is detected when using host-based SNMP monitoring and the Oracle ILOM trap proxy. You might want to add this option if you require more precise real-time data regarding sensor values (temperature/voltage). Set the default sleep time in seconds. Note - When using this option, Oracle recommends not setting it to anything under 60 seconds due to the additional use of system resources.
hwmgmtd_reset_hour=<`hour`>	To prevent possible memory leaks, the management agent service should be reset once daily. Choose a time when there is little server activity since restarting the agent temporarily increases I/O activity as it performs device discovery. Set the reset time in whole hours. Supported values are: 0-23. The default is 3 (for 3AM local time).

Log Levels: Basic and Bit Flag Parameters

The following table lists the supported basic values for the hwagentd_log_levels section in the hwmgmtd.conf file.

Log Level	Messages Logged
ERROR	Any error messages generated by the Hardware Management Agent
WARNING	Any error and warning messages generated by the Hardware Management Agent
INFO	Any error and warning messages generated by the Hardware Management Agent and informative messages about normal functioning

Using bit flag parameters allows you to set the logging level with a finer level of granularity. The following table lists supported values.

Note - Oracle recommends that you use the basic logging levels. The bit flag parameters are for advanced troubleshooting.

Log Level	Bit Code	Messages Logged
EMERG	0x0001	Information about the system being unusable
ALARM	0x0002	Information about any immediate action that must be taken
CRIT	0x0004	Information related to the Hardware Management Agent either not starting or stopping because of critical conditions
ERROR	0x0008	Information about any error messages generated by the Hardware Management Agent
WARNING	0x0010	Information about any error and warning messages generated by the Hardware Management Agent
NOTICE	0x0020	Information related to normal functioning
INFO	0x0040	Information about any error and warning messages generated by the Hardware Management Agent and informative messages about normal functioning
DEBUG	0x0080	Verbose debug-level messages, useful in troubleshooting
TRACE	0x0100	Highly verbose debug-level messages, useful in troubleshooting

Note - levels DEBUG and TRACE generate a lot of detailed messages and are designed for troubleshooting. These levels are not recommended for production usage.

For example, when you want to set all logging levels between EMERG and NOTICE, the bit code values of all the required levels must be added and then converted to a decimal value. Referring to preceding table, the addition would be as follows:

0x0001 + 0x0002 + 0x0004 + 0x0008 + 0x0010 + 0x0020 = 0x003f

Converting 0x003f to decimal equals 63, which is the desired log level. This is the decimal number that should be assigned to the hwagentd_log_levels parameter in the hwmgmtd.conf file.