Using SNMP

About SNMP

You use SNMP (Simple Network Management Protocol) to direct a process, called an agent, on the filer to perform network management tasks such as gathering status and diagnostic information. The information is sent to network management stations, which are client workstations on a network. The network management stations use third-party applications to process the information.

The information that is exchanged to perform these tasks is described in ASCII files called Management Information Bases (MIBs).

Data SNMP provides

For diagnostic and other network management services, the filer supports the SNMP MIB-II specification. Based on SNMP version 1, this specification provides data about the following MIB-II groups:

SNMP commands enable users to specify up to eight communities and trap notifications for up to eight management stations.

Command to configure the SNMP agent

To use SNMP, configure the SNMP agent using the snmp command. A typical set of SNMP commands in the /etc/rc file in the root volume is as follows:

snmp contact 'jdoe@abc.com 415-555-1212'
snmp location 'ABC corporation, engineering lab'
snmp community add ro private
snmp traphost add snmp-mgr1
snmp init

SNMP commands supported by Network Appliance

SNMP commands that Network Appliance supports are described in detail in the na_snmp(1) man page. The following paragraphs provide brief explanations of these commands:

About the Network Appliance custom MIB

The Network Appliance custom MIB provides detailed information about many aspects of filer operation. The current version of the Network Appliance custom MIB is 1.1.2 It contains objects that help you manage multivolume features.

Where to get the MIB

You can obtain the custom MIB from the Data ONTAP CD, or from the NOW Web page at http://www.now.netapp.com. To locate the MIB on the CD, read the intro.txt file.

Installing the MIB

Install the MIB file on your network management workstation according to the installation procedure for your workstation, so that your workstation can obtain information from the filer about the objects that are part of the MIB.

Using deprecated single-volume objects

Single-volume objects are deprecated, but you can use them for single-volume systems as before. If you are using a single-volume filer, you do not need to make changes to use the new MIB.

Finding multivolume objects

The descriptions of the deprecated single-volume objects contain the names of the corresponding new multivolume objects. For multivolume objects, use the new objects rather than the deprecated ones.

About MIB group contents

The top-level groups in the custom MIB and the information they contain are described in the following table
Group name Contents
cifs Statistics like those displayed by the cifs stat command.
filesys Information related to the file system, including the equivalent of the maxfiles and df commands, and some of the information from the snap list command. For more information about maxfiles, see "Increasing the maximum number of files." For more information about the df command, see "The df command." For more information about snapshots, see Chapter 11, "Snapshots."
nfs Statistics like those displayed by the nfsstat command, including statistics for each client if per-client statistics are enabled. The per-client statistics are indexed by client IP addresses.
product Product-level information, such as the software version string and the system id.
quota Information related to disk quotas, including the output of the quota report command. To access quota information, quotas must be turned On. For more information about quotas, see "Restricting or tracking disk usage by using disk quotas."
raid Information about raid equivalents to the sysconfig -r output. For more information about sysconfig, see Chapter 20, "System Information and Performance."
sysstat System-level statistics, such as cpu uptime, idle time, and the number of kilobytes transmitted and received on all network interfaces.
.

About traps

You use the snmp traps command to inspect the value of MIB variables periodically and send an SNMP trap to the machines on the traphost list whenever that value meets the conditions you specify. The traphost list specifies network management stations that receive trap information.

You use the SNMP third-party applications on your network management station to process the trap information. For example, you can set a trap to monitor the fans on a filer and have the SNMP application on your network management station put a flashing message on your console that tells you that a fan has ceased operating.

You can set traps on any numeric variable in the filer's MIB.

How to define traps

You define or change a user-specified trap using the following snmp traps command:

traps trapname.parameter value

trapname is the name of the trap.

value is the value that you assign to the definition.

parameter must be a parameter listed in the following table.

Parameter Description
var
A trap's variable is the MIB variable that is queried to determine the trap's value. All MIB variables must be specified in the form snmp.oid, where oid is an OID (Object Identifier). A list of OIDs in the NetApp MIB is in the traps.dat file in the same directory as the MIB.
trigger
A trap's trigger is a piece of code that determines whether the trap should send data. The following triggers are available:
  • single-edge-trigger--sends data when the trap's target MIB variable's value crosses an edge--a value that you specify.
  • double-edge-trigger--enables you to have the trap send data when an edge is crossed in either direction (the edges can be different for each direction.
  • level-trigger--sends data whenever the trap's value exceeds a certain level.
edge-1
edge-2
A trap's edges are the threshold values that are compared against during evaluation to determine whether to send data.
edge-1-direction 
edge-2-direction
Edge-triggered traps only send data when the edges are crossed in one direction. By default this is UP for the first edge and DOWN for the second edge. The direction arguments let you change this default.
interval
The interval is the number of seconds between evaluation of the trap. A trap can only send data as often as it is evaluated.
interval-offset
The interval offset is the amount of time in seconds until the first trap evaluation, and is zero by default. You can set it to a non-zero value to prevent too many traps from being evaluated at once (at system startup, for example).
backoff-calculator
After a trap sends data, you might not want it to be evaluated so often anymore. For instance, you might want to know within a minute of when a file system is full, but only want to be notified every hour that it is still full. There are two kinds of backoff calculators: stepwise and exponential.
backoff-step
The number of seconds to increase the evaluation interval if you are using a step backoff. If a trap's interval is 10 and its backoff-step is 3590, the trap is evaluated every 10 seconds until it sends data, and once an hour thereafter.
backoff-multiplier
The value by which to multiply a trap's evaluation interval each time it fires. If you set the backoff calculator to exponential-backoff and the backoff multiplier to 2, the interval doubles each time the trap fires.

It is important to distinguish between the kind of user-specified traps you can set using the snmp traps command, and the built-in support for traps defined in RFC 1213, such as cold-start. Built-in traps such as cold start are automatically sent to the hosts on the traphosts list when some event (a reboot in the case of a cold start) occurs. User-specified traps only exist after they are defined by a series of snmp traps commands.

Traps are persistent. After you set a trap, it remains across reboots until you specifically remove it.

Where to get further information

Refer to the following RFCs (Requests For Comments) for more information:


Copyright © 1999, Network Appliance, Inc. All rights reserved.

Part No. 210-02334
File: net2.htm
Last updated: 04/26/99 08:11:18