Administration Guide

Enabling network management for the SP system

For an SP to be part of a network management system, network managers must be notified of the AIX Error Log or Event Management events that you would like to centrally manage. A section of the SP proxy agent MIB (contained in the /usr/lpp/ssp/config/snmp_proxy/ibmSPMIB.my file) contains SP-specific SNMP configuration data, enabling Managers to view the SP as a system composed of processor nodes and a control workstation. In other words, there must be a way to create SNMP traps for SP error or problem events and there must be a MIB that contains configuration information specific to the SP system.

SNMP traps for the SP from AIX error log events

The snmp_trap_gen error notification method notifies the sp_configd daemon (or SP proxy agent) to create SNMP traps when selected types of errors are recorded in the AIX Error Log. (Note that the trapgend method from NetView for AIX also creates SNMP traps from AIX Error Log events. trapgend formats such events to be specifically compatible with SNA alerts. Only one of these methods should be used to avoid duplicate traps being generated for a single error.) The method and daemon run on each processor node and control workstation composing the SP system. In order for a trap to be created, using the snmp_trap_gen method, error log entries on each node and control workstation must contain an "alert=true" attribute. The following steps describe the flow of events for creating an SNMP trap for an AIX Error Log event:

An application or subsystem writes to the AIX Error Log facility.
If the Error Log entry contains an "alert=true" attribute, the snmp_trap_gen error notification method runs.
The snmp_trap_gen method uses the sequence number from the Error Log entry as input to the errpt -a command to obtain full information about the event.
snmp_trap_gen places the event information in a FIFO file, /var/tmp/errlog_entry
sp_configd reads the FIFO file, parses the information into objects within the ibmSPErrlogVars group of the ibmSP MIB and creates a trap from the objects whose instantiations contain non-null values.
sp_configd sends the trap to the AIX agent, snmpd, which in turn sends the trap to network managers specified in the /etc/snmpd.conf file existing on the node.

The following information is provided in SNMP traps from AIX Error Log events generated using the snmp_trap_gen method and the sp_configd daemon.

The enterprise field contains the OID (object identifier) of the sp_configd subagent (this is the OID assigned to the ibmSP MIB)
The specific-trap field contains the error ID from the Error Log entry (this is the convention expected by NetView for AIX when configuring events)
When the corresponding information exists in the Error Log entry, the variable bindings field contains the following object values paired with their OIDs:
- Error label
- Error ID
- Error log entry time stamp
- Unique sequence number
- Machine ID parameter
- Node ID parameter
- Error class
- Error type
- Resource name
- Resource class
- Resource type
- Identifies the location code of a device
- Vital product data
- Error description
- Probable causes
- User causes
- User actions
- Install causes
- Install actions
- Failure causes
- Failure actions
- Detail data

Other fields in the trap are as specified by SNMP protocol.

Note: the values for objects with a syntax of 'DisplayString' are always in the form of English ASCII character strings which are not subject to translation.

Designating AIX error log entries as alertable

In order for sp_configd to pass any entries in the system error log as traps to the snmpd daemon for transmission to network managers, the error record templates for the entries must have the specific "alert" field set to the value true. (Make sure you are authenticated as the root user.) This can be accomplished as follows:

On the system in which the sp_configd daemon is running, use the errpt -t command to list the templates for entries in the AIX Error Log.
Select the IDs of the entries for which you wish to have traps issued when they occur.
Create a file containing the following two lines for each entry to be designated alertable:
```
= ID:
alert = true
```
From the command line, enter:
```
errupdate file
```
To verify that the alert status has been changed to "true" (or 1), enter:
```
errpt -tF alert=1 | grep ID
```
listing the error template. You should see the specified ID listed.

SNMP traps for the SP from Event Management Events

The following steps describe the flow of events for creating an SNMP trap for an Event Management event:

Designating that an SNMP trap is issued for an Event Management event

Use the pmandef command to subscribe to an Event Management event and specify that an SNMP trap be issued for that event. For example:
```
pmandef -s Filesystem_Monitor /
-e'IBM.PSSP.aixos.FS%totused:NodeNum=10;VG=myvg;LV=mylv:X>95' /
-t 1234 -n 0
```
(In this example, whenever the file system associated with the mylv logical volume and myvg volume group on node 10 becomes more that 95% full, an SNMP trap will be generated on the control workstation.)
pmand (the Problem Management Event Management client daemon) subscribes to the event as specified in the pmandef command. The event must occur on the local node on which pmand is running.
pmand writes the contents of the Event Manager subsystem-supplied event response and the user-specified event configuration information into a FIFO file
sp_configd reads the data and creates an SNMP trap from it
sp_configd sends the trap to the SNMP managers specified in the /etc/snmpd.conf file on the node.
You specify a particular trap ID in the configuration information for an Event Manager event.

The following information is provided in SNMP traps from Event Management events generated by the sp_configd daemon.

The contents of SNMP traps containing Event Management events

The enterprise field contains the OID of the ibmSP MIB
The specific-trap field contains the trap ID field from the pmand configuration for this event. This allows configuration of NetView for AIX actions based on the type of event (which is done based on the specific trap ID)
The variable bindings field contains the following object values paired with their OIDs:
- The event ID
- Event flags
- A time stamp indicating the time the Event Manager generated the event
- The number of the node and the system partition address where the event occurred
- The resource variable name
- MIB table and instance information allowing further information about the variable involved with the event to be obtained from the ibmSP MIB.
- The value of the resource variable that triggered the event
- The predicate for which the event was triggered

Other fields in the trap are as specified by SNMP protocol.

The MIB for the SP

The ibmSP MIB is provided to define SP-specific information. The ibmSP MIB consists of an ibmSPConfig group, the ibmSPErrlogVars group, and the ibmSPEMVariables group. The ibmSP MIB is instantiated on each SP processor node and the control workstation.

The ibmSP MIB is defined in the source file ibmSPMIB.my in the /usr/lpp/ssp/config/snmp_proxy directory. For a sample of the ibmSP MIB file, see ibmSPMIB.my.

The ibmSPConfig group defines objects containing SP system configuration information. The group consists of objects containing the following information:

The node number of the node on which the queried subagent is running
The IP address of the system partition in which the queried node resides
The host name of the primary control workstation.
The operational state of the primary control workstation (up or down)
The host name of the backup control workstation (will be NULL if none exists)
The operational state of the backup control workstation (up or down)
The version number of the IBM Parallel System Support Programs for AIX running on the control workstation
A table of SPNodeEntrys. Each row in the table is indexed by the IP address of a system partition combined with a node number. Each row contains the following information about a node:
- The IP address of the system partition in which it resides
- The node number
- The number of the frame containing the node
- The lowest slot number in the frame containing the node
- The number of slots in the frame occupied by the node
- The reliable host name assigned to the node (this is the host name associated with the SP Ethernet)
- The initial host name assigned to the node (this is the host name assigned to the node during the customization based on data obtained from the SP Perspectives GUI)
- The name of the system partition in which the node resides
- The version number of the IBM Parallel System Support Programs for AIX running on the node

The ibmSPConfig group, when instantiated on the control workstation, will identify the default system partition IP address as the IP address of the system partition containing the reporting node and 0 as the node number of the reporting node.

The ibmSPErrlogVars group is instantiated on each SP processor node and the control workstation. The group consists of a sequence of objects containing information about the latest Error Log write that caused an SNMP trap to be issued on the reporting node.

The ibmSPEMVariables group is instantiated on each SP processor node and the control workstation. The group consists of a sequence of objects containing information about the last Event Management event that caused an SNMP trap to be issued on the reporting node and tables of objects representing Event Management variables. The group contains up to three tables:

A table of objects (ibmSPEMNodeDepVarsTable) allowing access to the definitions of node-dependent resource variables (those defined with an Event Manager Locator attribute). This table contains object instantiations for only those variables which have a Locator value equalling the node number of the node on which the SP proxy agent is running.
A table of objects (ibmSPEMNodeIndepVarsTable) allowing access to the definitions of node-independent resource variables (those defined without an Event Manager Locator attribute)
A table of objects (ibmSPEMVarValuesTable) allowing access to the current values assigned to Event Management resource variables

The following information is provided through object instantiations for each of the Event Management resource variables in the ibmSPEMNodeDepVarsTable and ibmSPEMNodeIndepVarsTable tables:

Strings containing the following Event Management variable information:
- Name of the resource variable
- Resource variable description
- Variable type (counter, quantity, or state)
- Data type (long, float, structured byte string)
- Structure byte string format (only for a structured byte string data type)
- Initial variable value
- An index into the ibmSPEMVarValuesTable, which, when used in combination with an instantiation vector, provides access to an entry containing the current value of a variable instance
- Variable's resource class
- Instantiation vector definition
- Description of each element in the instantiation vector
- Name used to read and write the variable in the PTX shared (may be NULL) memory
- Default predicate for event notification (may be NULL)
- A description of the event
- Locator value specified for the variable
- Resource order group

The ibmSPEMVarValuesTable provides object instantiations for the current values of variables from both the ibmSPEMNodeDepVarsTable and the ibmSPEMNodeIndepVarsTable tables.

The portion of the ibmSPEMVariables group instantiated by the sp_configd subagent depends on the type of SP node the subagent is running on. The ibmSPEMNodeDepVarsTable and ibmSPEMVarValuesTable tables are instantiated on all the processor nodes and the control workstation. The ibmSPEMNodeIndepVarsTable table is instantiated on only the control workstation.

Viewing the ibmSP MIB

The AIX snmpinfo command may be used in place of a manager client, such as NetView for AIX, to view the contents of the ibmSP MIB. Use of this command requires root authentication. The following are examples of using snmpinfo:

To dump the contents of the ibmSP MIB located on the host host_name in verbose mode, enter the command:
```
snmpinfo -m dump -v -h host_name ibmSP
```
To dump the contents of the ibmSPConfig group, enter the command:
```
snmpinfo -m dump -v -h host_name ibmSPconfig
```
To get values for the MIB variable SPhostnodenumber.0 located on the local host, enter:
```
snmpinfo -m get ibmSPhostnodenumber.0
```
To get the value for the MIB variable following the ibmSPhostnodenumber variable, enter:
```
snmpinfo -m next ibmSPhostnodenumber.0
```

[ Top of Page | Previous Page | Next Page | Table of Contents | Index ]