Method for diagnosing forwarding faults of network processor

Abstract

Using features of multithreading design in network processor, the disclosed method triggers diagnostic process based on diagnosis conditions input from users. In diagnosis, diagnosis thread controls diagnosed thread operated in single step, and records forwarding information in each single step operation. Reading the forwarding information carries out diagnosis. The method does not need to setup breakpoint in forwarding route. Through network processor setup remotely, facility maintenance personnel realizes automatic diagnosis of forwarding faults. Advantages are: raising speed for solving forwarding faults of network processor, and saving traveling expenses.

Description

technical field [0001] The invention relates to network processor technology, in particular to a method for diagnosing network processor forwarding faults. Background technique [0002] A network processor is a part of a network device that processes packets at a high speed. It generally contains multiple on-chip processors, and these on-chip processors are also called forwarding engines. Each forwarding engine may contain multiple threads, and a thread is a functional unit for forwarding and processing packets. In order to speed up the processing speed of the message, multiple threads can process data in parallel, which is a remarkable design feature of the network processor. Usually there is also a general-purpose processor in the network device. The role of this processor is to manage the network processor, calculate, generate and send forwarding entries to the network processor through the network protocol, and at the same time provide a The interface that interacts wi...

AI Technical Summary

Problems solved by technology

[0006] The above-mentioned first dot counting method is very effective when the packet type is relatively single, but for the network equipment running on the network, when the forwarding traffic is very large, there are many types of forwarded packets, and the counting of each critical path counter In the case of increasing, the fault cannot be effectively located only by observing the count

The second single-step diagnosis method above is an important fault diagnosis method in the code debugging stage, but for running network devices, there are two obvious disadvantages in using single-step diagnosis: First, if only some of the devices running on the network appear If there is a fault, rather than the failure of the entire network-running device, it is generally not allowed to interrupt the service, so it is not allowed to set a breakpoint on the forwarding path in this case; secondly, this method cannot set a breakpoint according to the characteristics of the message. When the packet forwarding traffic is heavy, it is almost impossible to capture faulty packets for single-step diagnosis

Images