KR20000038701A - Method for duplicating error monitoring of distribution processor using internet control message protocol - Google Patents

Abstract

PURPOSE: A method for duplicating an error monitoring of a distribution processor using an internet control message protocol(ICMP) is provided to correctly monitor an operation of an active process and increase reliability of the error monitoring, by enabling a standby processor to perform the monitoring using an echo message and echo reply message of the ICMP as a heart-beat message. CONSTITUTION: A communication initializing environment is set for an active processor and standby processor(S1). The standby processor performs a function for processing an acceptation in order to monitor an operation state of the active processor, and sets an alarm signal so as to prevent an infinite standby state for performing the acceptation processing function(S2). If the acceptation processing function has been performed successfully(S3), the active processor performs a function for performing the error monitoring(S4-S6). If the execution of the acceptation processing function has failed(S3), a server of the standby processor directly communicates with the ICMP of the active processor in order to check operation state of the active processor again, and then performs a corresponding process according to the operation state thereof(S8-S12).

Description

Fault Monitoring Redundancy Method of Distributed Processing Unit Using Internet Control Message Protocol (ICPM)

The present invention relates to a failure monitoring duplication method of a distributed processing apparatus using the Internet Control Message Protocol (hereinafter referred to as ICMP), wherein a plurality of processors are connected to a redundant local area network (LAN) to connect a system. In the distributed processing unit that provides the service, the fault monitoring function is duplicated in an active / standby manner. When the active processor fails and the fault monitoring activity cannot be performed, the standby processor automatically detects it. It is a technology that takes over the fault monitoring function of the active processor and continuously performs the fault monitoring function, and when the active processor is detected to be restarted, stops the fault monitoring function performed by the standby processor and returns it to its original state. On how to improve reliability Will.

In general, each processor constituting the distributed processing unit needs to be aware of the operation status of the other processor to provide reliable service without loss of data in interprocessor communication even in the event of a failure. Redundancy of the fault monitoring function is essential.

This redundant type of fault monitoring mainly uses the active / standby method to double the fault monitoring function. For this purpose, since the first active processor does not know when a failure occurs, the active processor and the standby processor are the targets of the distributed processing unit. There must always be a consistent state of failure, process failure, processor failure, service failure, etc.).

To do this, the active processor must pass the collected / analyzed fault messages and the configuration table (a table representing the current state of the fault monitor target) to the standby processor in real time. That way, when an active processor fails, the standby processor can continue to perform fault monitoring.

Second, the standby processor must accurately and accurately monitor the active state of the active processor in real time.

If the standby processor incorrectly monitors the state of the active processor, fault monitoring processes may operate on both the active processor and the standby processor.

Under these conditions, the redundancy scheme currently in use employs a daemon process on the active processor and the standby processor to allow two application processes to send and receive messages and monitor status. However, since the counterpart application process responds, it not only wastes the resources of the processor but also causes the processor failure when the counterpart process is down.

In view of the above problems and requirements, the present invention provides an echo of ICMP protocol (hereinafter referred to as ICMP_ECHO) provided by a transmission control protocol / Internet protocol (TCP / IP) for a standby processor to accurately monitor the operating state of an active process. By using the message and the echo response (hereinafter referred to as ICMP_ECHOREPLY) message as a heart-beat message, the fault monitoring is performed to improve the fault monitoring reliability.

In other words, when the ICMP protocol is used, the ICMP protocol is not responded to the ICMP_ECHO message sent by the standby processor's application, but the ICMP protocol responds with the ICMP_ECHOREPLY message. The reliability of monitoring can be increased.

Herein, the ICMP_ECHO message and the ICMP_ECHOREPLY message are one of the ICMP protocol messages used by the processor or router that finds an error in the Internet protocol communication to notify the processor that sent the packet causing the error. For example, a heart-beat message is a message used to periodically check an operation state of a target.

BRIEF DESCRIPTION OF THE DRAWINGS The block diagram which shows schematic structure of the shared processing apparatus to which this invention is applied.

2 is a block diagram showing fault monitoring redundancy of an active / standby structure according to the present invention in a sharing processing apparatus.

Figure 3 is a flow chart illustrating a fault monitoring and its processing in the fault monitoring standby processor according to the present invention.

<Explanation of symbols for the main parts of the drawings>

1: Active processor for fault monitoring

11: standby processor for fault monitoring

2, 21, 3, 31: processor a: server

b: client

In order to achieve the above object, in the failure monitoring redundancy method comprising an active processor and a standby processor in the distributed processing device,

A first step of establishing an initial communication environment between the active processor and a standby processor;

A second step of performing an acceptance processing function to monitor an operation state of an active processor in the standby processor, and setting an alarm signal to prevent the acceptance processing function from falling into an infinite standby state;

If it is determined that the acceptance processing function is successfully performed and the successful execution is completed, the active processor is operating normally. If the standby processor is in operation, the active processor is in a failure state. In this case, the standby process stops the fault monitoring daemon processes running on the standby processor and executes the routine to collect / analyze fault data collected by the active processor.If the fault monitoring daemon process is not running, Since the active processor has been operating normally, a third process of performing only a fault message collection / analysis routine to match a shape between the active processor and the standby processor and maintaining the active task is performed by the active processor;

If the acceptance processing fails, the client of the standby processor communicates directly with the Internet Control Message Protocol (ICMP) of the active processor to recheck the operation state of the active processor, and then performs the processing according to the operation state of the active processor. Including a fourth process to perform.

The above objects, features and advantages will become apparent from the following detailed description taken in conjunction with the accompanying drawings. Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram showing a schematic configuration of a sharing processing apparatus to which the present invention is applied, and is a processor (B, B ', C, C') (2, 21, 3, 31) that is duplicated in a load sharing scheme to process a service. );

An active processor (A) 1 and a standby processor (A ') 11 that are responsible for fault monitoring;

It is provided with a redundant local area network (LAN) 4 connecting the respective processors.

Here, the processor (A) 1 is an active processor that performs a failure monitoring function, a processor that collects / analyzes a failure message generated by each processor constituting the distributed processing device, and the processor (A ') 11 is Active processor (A) is a processor waiting in standby to back up the function of the active processor (A) in case of failure.

FIG. 2 is a diagram illustrating a fault monitoring traversal in a normal situation in the shared processing apparatus configured as shown in FIG. 1, wherein the processors 2, 21, 3, and 31 that process various services have servers a in them. The client (b) is present inside the active processor for failure monitoring (1), and the server (a) and the client (b: operates only when an active processor fails) exist in the standby processor (11) for monitoring. do.

In the above state, the client (b) process operating in the fault-monitoring active processor (A) 1 periodically processes the processors B, B ', C, C' (2, 21, 3, 31) sequentially connect to the server (a) process operating in 3, 31 collects and analyzes the fault message, and then updates the shape table.

Then, the server (a) operating in the standby processor (A ') 11 is connected to transmit the collected fault message and the shape table to match the shape between the two processors (A, A') (1, 11). Let's do it.

If the shapes between the two processors (A, A ') (1, 11) do not match, the standby processor (A') 11 maintains an incorrect shape for the distributed processing unit when the active processor (A) 1 fails. This can seriously affect service processing.

The server (a) process operating in the processor (A ') 11 waiting on standby fails in the active processor (A) 1 when there is no connection request from the active processor (A) 1 for a predetermined time. The ICMP protocol is forwarded to the active processor's ICMP protocol and waits for an ICMP_ECHOREPLY message.

That is, the ICMP_ECHO message and the ICMP_ECHOREPLY message are used as a kind of heart-beat message to check the operation state of the active processor A.

In this state, if the active processor (A) 1 does not send a response message for the ICMP_ECHO message, the server (a) process of the standby standby processor (A ') is the active processor (A) for failure monitoring (1). A fault has occurred, and a standby monitoring daemon (A ') 11 operates a failure monitoring related daemon (daemon) process so that the standby processor (A') 11 performs a failure monitoring main task.

In this way, the fault monitoring function can be duplicated because the process of collecting / analyzing fault messages from the fault-monitoring active processor (A) 1 acts as a client (b), so that the server (a) process operates. Knowing only the IP address and port number of (B, B ', C, C') (2, 21, 3, 31), the client (b) process can run on any processor a) A failure message can be collected / analyzed by attempting to access the process.

For reference, in the present invention, the fault monitoring dedicated processor (A, A ') (1, 11) is installed for the fault monitoring function, but the fault monitoring function may be implemented in a processor for processing a service without using a dedicated processor. Do.

If the failure occurs in the failure monitoring active processor (A) 1 and the client (b) in the failure monitoring standby processor (A ') 11 to perform the failure monitoring function, the failure monitoring The difference from the client (b) operating in the active processor (A) (1) is that the failure message is collected and stored in the form of a file, when the processor (A) operates again and transmits it to the processor (A) details Is to manage.

Hereinafter, the role of the server a process in the standby monitoring processor (A ') 11 for performing the above-described failure monitoring switching operation will be described in detail with reference to FIG. 3.

The standby monitoring processor (A ') 11 monitors the operating state of the active monitoring processor (A) 1 while the client (b) of the active processor (A) 1 constitutes a distributed processing apparatus. Receives the fault message and shape table collected / analyzed from the processors 2, 21, 3, and 31, and operates the fault monitoring related daemon processes operating in the active processor A when the active processor A fails. The procedure for this purpose is to duplicate the monitoring function. First, the server (a) of the standby processor (A ') 11 is a heart-beat message for monitoring the operation of the active processor (A) (1). It is possible to send / receive the ICMP_ECHO message and the ICMP_ECHOREPLY message, which is a response message, and to execute the client (b) and the acceptance processing function of the active processor (A) in block mode. Socket communication It sets a path (step 1 (S1)).

Subsequently, the server (a) of the standby processor (A ') 11 calls a call for an acceptance processing function (accept ()) for receiving a connection request (1) of the client (b) in the active processor (A) (1). Before performing the operation, the alarm signal (SIGALRM: this signal is provided by the Unix system) is set to generate an alarm signal after a predetermined time, and then the acceptance processing function call is performed.

At this time, the time when the alarm signal is generated, the client (a) operating in the active processor (A) (1) performs a failure monitoring circuit routine for all the processors (2, 21, 3, 31) constituting the distributed processing unit As a parameter depending on the time to perform once, it doubles the time required to perform the fault monitoring circuit routine once (second step S2).

In this state, it is determined whether the acceptance processing function returns successfully, and if the acceptance processing function call succeeds within a certain time, it is considered that the active processor A 1 operates normally, and the standby processor A The server a in the '11' releases the set alarm signal (step S3).

Subsequently, the server (a) in the standby processor (A ') 11 determines whether a failure monitoring-related process is running on the standby processor (A') 11, and if the failure monitoring-related processes are running, the server (a) operates in the standby processor. Disables fault monitoring related processes. In this case, the acceptance processing function call is successful because it means that the active processor is returned to the normal state from the failure state (S4).

After performing the fourth step, the server process operating in the standby processor (A ') 11 receives a failure message transmitted by the client (a) of the active processor (A) 1 (S5) )).

Then, by distinguishing whether the received message is a null message or not, analyzing the received failure message if it is not a null message, updating the status table, and receiving a failure message again (step S5). If the received message is a null message and there is no error message anymore, the connection is released and the process is repeated from the second step S2 (sixth step S6).

On the other hand, if the server (a) in the standby processor (A ') 11 is blocked for more than a certain time in the acceptance processing function call in the third step (S3) and the timeout occurs, the set alarm signal is It is transmitted to the server process (a) and the acceptance processing function call fails (seventh step S7).

If the attempt to accept processing fails, the server (a) of the standby processor (A ') 11 uses the ICMP_ECHO message, which is used as a heart-beat message, to accurately determine the state of the active processor (A) (1). A) It repeats twice the routine that delivers the ICMP level of (1) and waits for the ICMP_ECHOREPLY message, which is a response message.

The method for repeating the routine is to initialize a variable (i = 0) indicating the number of times the routine is repeated and a variable (fail = 0) indicating the number of times the ICMP_ECHOREPLY message is not responded, and then fail value 2 when i becomes 2. If so, it is determined that the active processor has failed. In this case, the transmission of the ICMP_ECHO message twice is for accurate determination (S8).

In this case, since the ICMP_ECHOREPLY message cannot be waited indefinitely in step 8, the alarm signal is set to 1 second before performing the acceptance processing function call, and then the ICMP_ECHO message is transmitted (step S9).

Then, it is checked whether the ICMP_ECHOREPLY message has been received from the active processor (A) 1, and if the message is received, the alarm signal set to 1 second is released. In this case, the active processor (A) 1 is operating normally, and a client process running in the active processor (A) 1 has failed and connected to the server (a) of the standby processor (A ') 11. Since no request is made, a client process failure message of the active processor (A) 1 is generated (S10).

If the ICMP_ECHOREPLY message has not been received as a result of the message acknowledgment, the value counting that the message has not been received is increased by one (fail ++), and then whether the current message checking routine is repeated twice (i == 2?) If the repetition is not completed, the routine is repeated again, and if the repetition is completed, it is checked whether the message reception check has not been received twice (fail == 2?).

If the message is received at least once as a result of the check, an alarm signal release performed in the tenth step S10 and a client process failure occurrence message in the active processor A 1 are generated (an eleventh step S11).

On the contrary, if the ICMP_ECHOREPLY message is not received twice as a result of the message reception check, the fault monitoring related daemon process operating in the active processor (A) 1 is considered to be down. The standby processor (A ') 11 is started to cause the standby processor (A') 11 to perform the failure monitoring main task (12th step S12).

The reason why the server (a) in the standby processor (A ') 11 transfers the ICMP_ECHO message twice to the active processor (A) (1) is to accurately determine the operating state of the active processor (A).

The server of the standby processor (A ') 11 receives a connection request from the client (b) of the failed active processor (A) 1 while the processor (A') 11 performs the main task of fault monitoring. Keep an eye on it.

If the connection request from the client (b) of the active processor (A) 1 succeeds again, since the active processor (A) 1 is driven again, the standby processor (A ') 11 performs the main task of fault monitoring. By stopping the daemon processes that are running, the active processor (A) 1 controls to perform the main task of fault monitoring again.

As described in detail above, the present invention duplicates the failure monitoring function of the distributed processing device in an active / standby manner, and when a failure occurs in the active processor, the standby processor detects this and automatically standbys the failure monitoring daemon processes operating in the active processor. It improves the reliability of distributed processing device by redundancy by operating on the network and heart-beats ICMP_ECHO message and ICMP_ECHOREPLY message of ICMP protocol by the standby processor to monitor the operation status of active processor in active / standby device. By using it as a message, it entails the advantage of improving the accuracy of the active processor operating state monitoring operation of the standby processor.

In addition, preferred embodiments of the present invention are disclosed for the purpose of illustration, those skilled in the art will be able to various modifications, changes, additions, etc. within the spirit and scope of the present invention, such modifications and modifications belong to the following claims You will have to look.

Claims (8)

A failure monitoring redundancy method comprising an active processor and a standby processor in a distributed processing apparatus, A first step of establishing an initial communication environment between the active processor and a standby processor; A second step of performing an acceptance processing function to monitor an operation state of an active processor in the standby processor and setting an alarm signal to prevent an infinite waiting state of performing the acceptance processing function; A third step of determining whether to perform the acceptance processing function successfully, and if the successful execution is completed, performing a failure monitoring function in an active processor; If the acceptance processing fails, the server of the standby processor directly communicates with the Internet Control Message Protocol (ICMP) of the active processor to recheck the operation state of the active processor, and then performs the processing according to the operation state of the active processor. A failure monitoring redundancy method of a distributed processing apparatus using the Internet Control Message Protocol (ICMP), characterized in that performed by the fourth process. The method of claim 1, The first process sets an environment in which a server of a standby processor transmits and receives messages transmitted and received in order to monitor the operation of an active processor, and performs socket communication with a client of the active processor in block mode. A failure monitoring redundancy method of a distributed processing device using the Internet Control Message Protocol (ICMP), characterized in that the setting. The method of claim 1, In the second process, the alarm signal is set to twice the time required for the client operating in the active processor to perform the failure monitoring circuit routine once for all the distributed processing devices constituting the distributed processing device. Fault monitoring redundancy method of distributed processing apparatus using Internet Control Message Protocol (ICMP). The method of claim 1, The third process includes the first step of releasing the set alarm signal when the acceptance processing function is successfully completed; A second step of determining whether the standby processor is performing a failure monitoring function by the server process in the current standby processor and stopping the operation if the failure monitoring function is being performed; Determining that the standby processor is not performing the failure monitoring function, stopping the failure monitoring function of the standby processor, and receiving failure data transmitted by a client of the active processor; A fourth step of checking whether the received message is a null message and if it is not a null message, analyzing the received failure message and updating a state table to maintain a shape consistent with an active processor, and then performing a failure message reception routine again; If the received message is a null message, the connection is released using the Internet Control Message Protocol (ICMP). Failure monitoring redundancy method of processing unit. The method of claim 1, In the fourth process, the client of the standby processor is a heart-beat message for directly communicating with the active processor Internet Control Message Protocol (ICMP), and an ICMP_Echo (ECHO) message and a response message thereof. ICMP_ECHOREPLY) is a failure monitoring redundancy method of the distributed processing device using the Internet Control Message Protocol (ICMP). The method of claim 5, When the ICMP echo signal is transmitted, an alarm signal is set in order to prevent an infinite waiting for response reception. The failure monitoring redundancy method of the distributed processing apparatus using the Internet Control Message Protocol (ICMP). The method of claim 1, In step 4, the process of receiving the ICMP_response message for analyzing the operation state of the active processor is repeated twice. If the message is received more than once, a failure message indicating that a failure occurs only in the client process in the active processor is repeated. The generation and failure monitoring redundancy method of the distributed processing apparatus using the Internet Control Message Protocol (ICMP), characterized in that it is repeated from the second process. The method of claim 7, wherein If the message is not received two times as a result of the repetition, it is regarded as a failure in the active processor, and the standby processor starts the failure monitoring daemon process operated in the active processor so that the standby processor performs the failure monitoring main task. Redundancy Monitoring of Fault Monitoring Devices Using Internet Control Message Protocol (ICMP)