9.1 Network Interface Failure Finder (NIFF)

9.1 Network Interface Failure Finder (NIFF)

The Network Interface Failure Finder is a new component to Tru64 UNIX that enables network interfaces to be monitored for connectivity issues. In order for a network interface to be monitored on a Tru64 UNIX standalone environment, you need to inform NIFF using the niffconfig (8) command.

Note

In a TruCluster Server environment, configured network interfaces are automatically monitored by NIFF.

Once an interface is added via niffconfig, the kernel-level portion of NIFF, called the Traffic Monitor Thread (nifftmt(7)), monitors the interface. If the interface fails to respond, nifftmt posts an event to EVM.

The user-level NIFF daemon (niffd (8)) subscribes to the EVM events generated by nifftmt. When a connectivity issue arises, it is niffd's job to generate traffic on the interface that has been marked inactive by nifftmt to verify whether the interface is truly down.

NIFF is used in conjunction with NetRAIN (see below) to provide a high-availability network solution for a standalone system. In a cluster, NIFF is also used by the Cluster Alias (CLUA) subsystem as well as the Cluster Application Availability (CAA) subsystem. We will cover CLUA and CAA in more detail in chapter 16 and chapter 23 respectively.

To add an interface to be monitored by NIFF, type the following command (replace tu0 with the interface you wish to monitor).

 # niffconfig –a tu0 

To examine the state of the interface according to NIFF, you can use either the "-u" or "-v" switches as shown in the following examples:

 # niffconfig –u Interface: tu0, status: UP 

 # niffconfig –v Interface: tu0, description:     , status: UP, event: ALERT, state: GREEN t1: 20, dt: 5, t2: 60, time to dead: 30, current_interval: 20, next time: 20 

Figure 9-1 shows the nifftmt to niffd communication path.

Figure 9-1: NIFF Flow

The nifftmt thread checks the monitored interface every t1 seconds. If nifftmt detects a connectivity problem, then it posts a sys.unix.hw.net.niff.alert event to EVM and then starts checking the monitored interface every dt seconds. The niffd daemon receives the sys.unix.hw.net.niff.alert event from EVM and begins generating traffic on the interface until it receives a sys.unix.hw.net.niff.down or sys.unix.hw.net.niff.up event. If t2 seconds elapse without the monitored interface's counters increasing, then the interface is marked as DEAD, and a sys.unix.hw.net.niff.down event is sent by nifftmt.

9.1.1 NIFF Events

NIFF events are defined in /usr/share/evm/templates/sys/hw.net.evt. Below are the registered NIFF events:

 # evmwatch -A -i -f "[name *.niff]" -t "[@priority] @name" [600] sys.unix.hw.net.niff.down [200] sys.unix.hw.net.niff.up [200] sys.unix.hw.net.niff.alert 

For additional information regarding NIFF, see the Tru64 UNIX Network Administration Guide for version 5.0A and version 5.1, or the Tru64 UNIX Network Administration: Connections Guide for version V5.1A. You might also check out the niff (7), nifftmt (7), niffd (8), and niffconfig (8) reference pages.