KR19990050461A - Error Handling in High Availability Systems - Google Patents

Abstract

The present invention relates to an error handling method of a high availability system, and when a failure occurs in an input / output board other than the system control board during operation of the high availability system, logic for detecting an error of the system control board and the input / output board is added. In order to minimize the downtime of the system, it is possible to prevent the service interruption caused by the malfunction occurring in the process of searching the status information of the counterpart system through the public network and the serial line.

The error handling method of the high availability system is that two systems having the same configuration share services with each other through an input / output process, and when an error occurs in any one system according to the execution of the error detection process. Delivering the generated error information to the high availability management processor; In addition to receiving error information, if an error occurs in one of the two input / output processes, an operation for bringing down the corresponding system where the error occurred by an input / output control command is performed and an error is not generated. If not, this is accomplished by continuing to perform error detection.

Description

Error Handling in High Availability Systems

The present invention relates to a high availability system, and more particularly, when a failure occurs in a board other than the system control board during a high availability system operation, the system control board is reset to reduce down time of the system. Error handling method of high availability system to minimize.

Generally, a computer system prevents the system from being stopped due to a failure caused by software, hardware, etc., and prepares a program to check whether the internal operation of the processor is operating in unstable state. It maintains completeness and improves the reliability, availability, and maintainability of the system by detecting failure sites early and automatically determining the failure status if a failure occurs.

In order to improve the performance of the system, technical linkage between hardware technology and software is required. In a redundant system, the status monitoring of the counterpart system is usually a SCSI (SCSI: Small Computer System Interface) bus for controlling network channels and system peripherals. This was done using a bus.

FIG. 1 is a schematic block diagram showing a high availability system according to the prior art, in which a first system 20, a second system 30, and an interface are connected to a public LAN as shown. The first system 20 and the second system 30 are interconnected through the cluster management system 10 and the serial line, and are composed of shared disks 40 which are shared.

Preferably, the first system 20 and the second system 30 are server / client processes 24, 26, 34, 36, error detection processes 28 and 38 for detecting errors, Cluster management daemon processes (CMSD: hereinafter referred to as daemon processes) 22, 32.

The error handling process of the high availability system configured as described above will be described in detail with reference to FIG.

First, when the first and second systems 20 and 30 are activated, a high availability setup process (HASETUP Process) is performed to synchronize with the remote system (ST10) (ST11).

The high availability setup process is performed in the first and second systems 20 and 30, respectively, in which a serial line connecting the mutual systems is connected to the SLIP (Heart Beat) to monitor the status of the counterpart system. Internet Protocol) (ST12) (ST13).

The high availability setup process then uses the serial line's state monitoring function to check if the counterpart system is ready, and if not, polling the counterpart until it is ready. That is, when inspecting the second system 30, the first system 20 periodically checks the state of the second system 30 until the system is activated through the serial line ( ST14) (ST15).

Subsequently, when the counterpart system becomes active, it is subsequently checked whether the counterpart system is active in a high availability state (ST16) (ST17). At this time, if the partner system is active in a high availability environment, after restoring resources of the other system, it waits for a command to complete the recovery, and starts status monitoring information of the partner system using the serial line (ST18) (ST19). (ST23).

If the counterpart system is not active in a high availability state, the first and second systems 20 and 30 are all initial states, so the HAM necessary to set up a public LAN and configure a high availability system is required. (High Availability Manager) process, fault detection process, and daemon process (ST20) (ST22).

That is, the high availability process performed by the high availability setup process periodically polls state information of the counterpart system using a remote procedure call (RPC) for calling the counterpart system to configure the high availability system. At this time, the path using the status information of the counterpart system is the method using the public network and the method using the serial line described above.

In this way, the status information of the other system is monitored. If the method of receiving the status information of the other system using the public network, which is the first method, fails (ST24), the other system using the serial line, the second method, is failed. State information is checked (ST26). If the status information check through the serial line also fails, the counterpart system has failed (ST28). In this case, the cluster management system 30 is inspected, and a task for providing a service provided by the counterpart system to another system is performed (ST30).

In order to provide such services, first, the client processes 26 and 36 of the first and second systems 20 and 30 must be continuously operated regardless of whether the server processes 24 and 34 have failed. In other words, the client processes 26 and 36 requesting the service need continuous operation, and thus provide the service by the Internet Protocol. The service provision of client processes 26 and 36 by the Internet protocol is set up to be the same Internet protocol address in order to be able to access the Internet protocol address of the same server process 24 and 34.

After that, the service provided by the failing system is transferred to the counterpart system and the system is continuously operated. When the service is completed, the service providing process is restored according to the recovery command, and the system is switched to the normal state. ST32 to ST38).

However, the error handling method of the high availability system according to the related art is a path for detecting a failure by periodically receiving information of the counterpart system, that is, monitoring the status of the counterpart system using a public network and of the counterpart system using a serial line. Since the status monitoring path is made through the system control board, if a failure occurs in a board other than the system control board and the system control board operates normally, the system may operate abnormally but in a normal state.

In this case, the client processes connected to the network are in a state where the service is interrupted. The failed system must continue to share the service with the counterpart system, but the system control board seems to be operating normally. I do not lose.

Accordingly, the present invention adds logic for detecting an error in the system control board and the input / output board when a failure occurs in an input / output board other than the system control board during the operation of the high availability system. By minimizing service downtime, the objective is to provide a method for handling errors in a high availability system that makes the system more stable.

1 is a schematic block diagram showing a high availability system according to the prior art,

2 is a flowchart illustrating an error processing process of the high availability system according to FIG. 1;

3 is a flowchart illustrating an error handling process of a high availability system according to the present invention.

<Explanation of symbols for main parts of the drawings>

10: cluster management system 20, 30: first, second system

22, 32: daemon process 24, 34: server process

26, 36: client process 28, 38: error detection process

40: shared disk

In order to achieve the above object, an error handling method of a high availability system according to the present invention includes two systems having the same configuration, each operating while sharing a service through an input / output process, depending on the execution of the error detection process. Transferring the generated error information to a high availability management processor when an error occurs in any one system; In addition to receiving the error information, if an error occurs in any one of the input / output processes of the system, an operation for bringing down the corresponding system where an error occurs by an input / output control command is performed, and an error is not generated. If not, it involves continuously performing error detection.

Preferably, the system down operation of the second process may include transferring the error information to a kernel function using an input / output control function; The kernel function receiving the error information includes resetting the system control board.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings.

Here, in describing the configuration of the embodiment of the present invention, the high availability system according to the prior art described at the beginning of the specification is also applied to the present invention, and may be omitted to avoid duplication of description, and the same reference numerals are used.

First, the high availability system performs a method using a public network and a method using a serial line to determine state information of the first and second systems 20 and 30. At this time, the initial setup process is performed to synchronize with the counterpart system at the start of one system.

That is, an operation for monitoring the counterpart system is continuously executed in the high availability system in operation, and information is provided through the public network and the serial line. If an error occurs during the provision of such information, the service detects it and provides the service contents of the failed system to the counterpart system so as to continuously execute the service.

In this series of operations, if the status monitoring of the counterpart system through the public network fails, the second method, the monitoring of the counterpart system via the serial line, is performed. It considers that it occurs and provides the service provided by the other system to another system.

In this case, if an error occurs in a board other than the system control board, it may not be detected and the system may operate as a normal state. In order to prevent this, a fault detection process as illustrated in FIG. By doing so, the status of the input / output process and the system control board can be identified to provide continuous service.

When the error detection process is performed (ST100), first, it is determined whether an error occurs in the input / output processor and the system control board, and when an error occurs, information about the error occurred in the high availability manager process is displayed. It delivers (ST102) (ST104) (ST106).

At this time, if the system control board operates normally and only the input / output process generates an error (ST108), the system does not share the service with the counterpart system because the system does not recognize the error caused by the abnormal operation. can not do it. Accordingly, the operation of resetting the system control board is performed to recognize that the inspection of the status information of the counterpart system is disabled through the public network and the serial line.

First, in a state in which error information detection through the I / O board is disabled, the system cannot be shut down using commands existing on the disk. So you have to take your system down by using processes and kernel functions in the system's memory.

The system down operation is handled by a high availability management process that resides in memory, and the high availability management process receives error information of an input / output board by an error detection process. Accordingly, the high availability management process transfers the reset command of the system control board to the kernel function by using an input / output control system call function of the kernel function (ST112).

The kernel function receiving the command resets the system control board by using a specific area of the system control board (ST114).

After that, when the system control board is reset, it is recognized that the method of checking the status information in the counterpart system, that is, the status information check through the public network and the serial line, fails, and the system in which the error has occurred is not able to share the service shared with the counterpart system. To provide (ST116).

Here, in the state where the input / output process does not generate an error, error detection is continuously performed by the error detecting process (ST110).

As described in detail above, the present invention adds logic for detecting an error of the system control board and the input / output board when an error occurs in an input / output board other than the system control board during operation of the high availability system. It is effective in minimizing downtime of the system by preventing service interruption due to a malfunction occurring while searching the status information of the counterpart system through the serial line.

Claims (2)

While two systems having the same configuration are sharing services with each other through input / output processes, if an error occurs in any one system according to the execution of the error detection process, the generated error information is transmitted to the high availability management processor. Conveying process; In addition to receiving the error information, if an error occurs in any one of the input / output processes of the system, an operation for bringing down the corresponding system where an error occurs by an input / output control command is performed, and an error is not generated. And if not, continuously performing error detection. The method of claim 1, wherein the system down operation comprises: transmitting the error information to a kernel function using an input / output control function; The kernel function receiving the error information comprises the step of resetting the system control board.