NOTE: To view a list of Frequently Asked Questions pertaining to SysFaultMgmt on HP-UX 11i v1, click here.
General questions:Problems
- How do I configure SysFaultMgmt?
- Which CIM Schema version is supported by the System Fault Management Providers?
- How can I find the definitions of the HP Extended MOF classes?
- Why are certain properties available on some platforms and not on others?
- Why are certain properties not reported by the instance providers?
- How do I use the EMS Wrapper Provider?
- How do I generate a sample indication?
- What versions of OnlineDiag are required for SysFaultMgmt?
- Do I need to load all of OnlineDiag? Can I just load the EMS-Core?
- How can I stop redundant notifications of events?
- When I select the SysFaultMgmt bundle by running swinstall in the interactive mode, I see that the selection is "Partial". Is this Okay?
- Events from certain monitors are not received by the Indication Consumer.
- Client requests time out on some systems.
General Questions
How do I configure SysFaultMgmt?
Currently only logging-related configuration can be changed for SystFaultMgmt. In general, HP recommends that you keep the default configurations for SysFaultMgmt and the providers as they come from the factory.
If desired, you can change the logging threshold or specify a new logfile.
To configure logging for SysFaultMgmt:
- With an ASCII editor, call up the configuration file /opt/sfm/conf/FMLoggerConfig.xml. The contents of the file should appear similar to the following:
<SFMConfig> <LoggerConfig> <Severity> WARNING </Severity> <Target> /opt/sfm/log/sfm.log </Target> </LoggerConfig> </SFMConfig>- To modify the threshold logging level, change the Severity value. In increasing level of severity, the possible values are:
INFORMATIONAL WARNING (default) ERROR CRITICAL
CAUTION
HP suggests that you not set the threshold to INFORMATIONAL, except perhaps temporarily. The INFORMATIONAL setting can generate many messages and use a sizable amount of disk space.
- To specify a new logfile, change the Target value. The possible values are: STDOUT (all log messages are delivered to console, where cimserver was started) or the complete path to the file where the log messages are to be written. The default logfile is /opt/sfm/log/sfm.log. (The path to the logfile should already exist and should be writable by root-user).
- After you have finished editing the configuration file, run the /opt/sfm/bin/fmControl program, to specify the changed configuration file. Specify the complete path of the configuration file. For example:
$ /opt/sfm/bin/fmControl /opt/sfm/conf/FMLoggerConfig.xmlWhich CIM Schema version is supported by the System Fault Management Providers?
The System Fault Management Providers support CIM Schema v2.7.2. The providers are not compatible with versions of the CIM Schema previous to v2.7.2.
How can I find the definitions of the HP Extended MOF classes?
The HP extended MOF classes are delivered in the /opt/sfm/schemas/mof directory. Please refer to the following files for more information:
- HP_MemoryPhysical27.mof : for memory-specific HP MOF extensions
- HPProcessor.mof : for processor-specific HP MOF extensions.
- HPDeviceIndiacation.mof : for indication-specific HP MOF extensions.
(MOF stands for Managed Object Format and is the language used to represent CIM classes.)
Why are certain properties available on some platforms and not on others?
Certain properties are not available on a few platforms, On those platforms, the CPU and Memory Instance Providers add a NULL CIM Value. For example, if the serial number/part number of a memory module is not available on a supported server, a NULL CIM Value will be added for that instance.
Why are certain properties not reported by the instance providers?
The instance providers don't add deprecated properties (= properties designated to become obsolete) to CIM instances. Hence, the SysFaultMgmt providers do not make these properties available in the instance.
How do I use the EMS Wrapper Provider?
Most customers will use the HP Systems Insight Manager (SIM) to subscribe to the EMS Wrapper Provider. See the SysFaultMgmt Operating Guide for instructions.
How do I generate a sample indication?
A sample indication can be generated by using the /etc/opt/resmon/lbin/send_test_event command provided by EMS. The syntax is:
/etc/opt/resmon/lbin/send_test_event MONITOR_NAMEwhere MONITOR_NAME is one of the monitors supported by EMSWrapper. For example, to send a test event to the Core Hardware Monitor (dm_core_hw), enter the command:
/etc/opt/resmon/lbin/send_test_event dm_core_hwThe monitor must be running for successful reception of the indication. To check whether the monitor is running use the following command:
ps -eaf | grep MONITOR_NAMEFor a complete procedure, see the Verifying EMS Hardware Monitors webpage.
What versions of OnlineDiag are required for SysFaultMgmt?
HP strongly recommends you install the latest version of OnlineDiag for your operating system to ensure you have up-to-date functionality for all devices. The minimum version for OnlineDiag on HP-UX 11i V2 is B.40.00, but newer versions may be required for some platforms.
(For more information on STM and EMS versions, see Finding your version number and Table of versions.)
Do I need to load all of OnlineDiag? Can I just load the EMS-Core?
If the server does not have OnlineDiag (with EMS-Core) already loaded, you must load the entire OnlineDiag bundle, not just the EMS-Core portion of it. SysFaultMgmt requires the EMS hardware monitors that are included in the OnlineDiag bundle in addition to the EMS infrastructure (EMS-Core).
How can I stop redundant notifications of events?
SysFaultMgmt provides one method of reporting hardware events (indications), and the EMS hardware monitors provide another method. Note, however, that SysFaultMgmt currently only supports a sub-set of all the available EMS hardware monitors. Also, note that all the various targets for events that the EMS monitors currently support (like tcp) may not be available from HP Systems Insight Manager.
You can use the Hardware Monitoring Request Manager to avoid redundant notifications on selected monitors. On the server with the EMS hardware monitors installed, enter the command: /etc/opt/resmon/lbin/monconfig. For each monitor that generates redundant notifications, use the Modify Monitoring Request (M) command and eliminate the duplicate notification method.
In this way, the EMS hardware monitors continue to run (as they should) but duplicate notification is avoided. If you have ISEE functionality, it is maintained.
For more information, see the EMS Hardware Monitor web pages at http://docs.hp.com/hpux/diag/index.html
When I select the SysFaultMgmt bundle by running swinstall in the interactive mode, I see that the selection is "Partial". Is this Okay?
For example, when you enter swinstall's interactive mode by entering "swinstall -s /tmp/depot_file", instead of by entering "swinstall -s /tmp/depot_file SysFaultMgmt", you see that the selection is "Partial".
This is perfectly alright. The SysFaultMgmt bundle contains filesets for all platforms in the same depot file. When a manual selection is made, filesets related to other platforms are not selected during installation.
Problems
Events from certain monitors are not received by the Indication Consumer
First, check if the monitor which should be generating the event is supported by the EMS Wrapper Provider. (The latest information about supported hardware monitors is in the System Fault Manager Release Notes.)
If the monitor is supported and the indications are still not being received, check if the monitor is active. Certain monitors run only on specific platforms. To check whether the monitors is active or not, use the resls RESOURCE_PATH command. For example, to see if the LPMC Monitor (lpmc_em) is active, enter:
resls /system/events/cpu/lpmcIf the monitor is not active, a message with the following text is displayed: system/events/cpu/lpmc is not a valid resource name.
The supported monitors and the resource paths (as of September 2004) are listed below:
For PA-RISC servers:
- LPMC (now CPU) (lpmc_em): /system/events/cpu/lpmc
- Memory (dm_memory): /system/events/memory
- Core HW (dm_core_hw): /system/events/core_hw
- Chassis Code (dm_chassis): /system/events/chassis
- Disk (disk_em): /storage/events/disks/default
- High Availability Disk Array (ha_disk_array): /storage/events/disk_arrays/High_Availability
- Integrity Core Hardware Monitor(ia64_corehw): /system/events/ia64_corehw
- IPMI Forward Progress Log Monitor (fpl_em): /system/events/ipmi_fpl
For IPF servers:
- Corrected Platform Error Monitor (cpe_em): /system/events/cpe
- IPMI Forward Progress Log Monitor (fpl_em): /system/events/ipmi_fpl
- CMC Monitor (cmc_em): /system/events/cpu/cmc
- Itanium Core Hardware Monitor(ia64_corehw): /system/events/ia64_corehw
- SCSI Disk Monitor (disk_em): /storage/events/disks/default
- High Availability Disk Array Monitor (ha_disk_array): /storage/events/disk_arrays/High_Availability
- Itanium Memory Monitor (memory_ia64): /system/events/memory_ia64
If the resource is active and the consumer is still not receiving the indication, repeat the /etc/opt/resmon/lbin/send_test_event process after 5 minutes. The EMS Wrapper polls for monitors that have become active after a pre-defined polling interval. If the monitor that is generating events has come up between the poll periods or after the cimserver is started (for example, at the time of reboot) then the registration for the monitor is added on the next poll.
Client requests time out on some systems.
On certain systems (typically, cellular systems), the providers may take a considerable amount of time to fetch all the required information, and the first request may time out. In such situations, increase the client timeout intervals (for example, on HP Systems Insight Manager). Longer timeout values might be required for higher-end systems. Responses to subsequent requests will be much faster. Also, retries must be made in quick succession to see the information.