JP2012173752A - Cluster system, data recording method, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent an error from occurring at the time of a switch from an active server to a standby server, by correctly recording data in a log file on a common disk.SOLUTION: A cluster system is so configured that a server device 10a in an active system and a server device 10b in a standby system perform with the use of a common disk 30. When a log message of an application 21 in a cluster service 20 is temporarily stored to a named pipe 13a on a local disk by a syslogd 11a having a data transfer function in the server device 10a, the log message stored in the named pipe 13a is recorded in a log file 31 on the common disk 30 by the log application 14a. In this process, while a log message of an application 21 is stored to the named pipe 13a, the log application 14a performs open/write/close processing on the log file 31.

Description

  The present invention relates to a cluster system that uses a shared disk by cooperatively configuring an active server and a standby server, a method for recording data in a log file on a shared disk in the cluster system, and a program.

  In general, in an active-standby type (active system and standby system) cluster system using a shared disk, when syslogd having a function of transferring a log message records a log message in a log file, the following is performed. It is done in a simple procedure. That is, when a server is operating as the active side (active system) of a cluster system, when an application on the cluster service in that server outputs a message to syslogd, that syslogd reads the message and writes the log file. Open a file in mode. Syslogd outputs the read message to the log file, and closes the log file when the output of the message is completed.

  FIG. 6 is a block diagram showing a configuration of a general Active-Standby type cluster system. In FIG. 6, when the server apparatus 110a activates the application 121 that implements the function of the cluster service 120, when the application 121 sends a message to the syslogd 111a having the transfer function, the syslogd 111a sends the message to a log file on the local disk. To 112a and 113a.

  Next, when the device that executes the cluster service 120 is switched from the server device 110a to the server device 110b, the application 121 that implements the function of the cluster service 120 is transferred to the log file 112b via the syslogd 111b on the server device 110b. , 113b. In this way, since messages related to the operation of the application 121 that implements the functions of the cluster service 120 are managed by a plurality of servers (two servers, the server device 110a and the server device 110b in FIG. 6), the log files are aggregated. It is necessary to merge (integrate) or compare a plurality of log files (for example, the log file 112a and the log file 112b) in a scene such as a trouble investigation.

  As a related technology, when a failure occurs in the active server in a system including the active server and the spare server, the operation is continuously switched to the spare server based on the status information recorded on the shared disk. A technology of a hot standby system is disclosed (for example, see Patent Document 1). According to this technique, information up to the time when a failure occurs in the active server can be automatically reproduced in the spare server. In addition, when a failure occurs, the active server and the spare server can be switched alternately.

  As another related technique, log data output from an application is stored in a named pipe corresponding to a memory-based queue (that is, FIFO: First In First Out) that is confirmed using a file name. Discloses a technique for remotely monitoring output data of an application program (see, for example, Patent Document 2). According to this technique, by reading the queue by opening the named pipe, the output data can be searched and the output data of the application program can be monitored remotely.

JP 11-259326 A Special table 2009-519509 gazette

  Generally, the log message transfer syslogd used as a log output daemon (software that provides various services resident in memory) in UNIX (registered trademark) or Linux (registered trademark) When a log file on the shared disk is specified at the transfer destination to be output, the problem described with reference to FIG. 7 occurs.

  FIG. 7 is a block diagram showing a configuration when a general Active-Standby type cluster system designates a log file on a shared disk. In FIG. 7, when two servers constituting the Active-Standby type cluster system (that is, the active server device 210a and the standby server device 210b) are activated, log messages belonging to the respective server devices are transferred. Syslogd 211a and Syslogd 211b start up and attempt to open the log file 231 on the shared disk 230. However, neither the server apparatus 210a nor the server apparatus 210b mounts (recognizes) an area where the log file 231 on the shared disk 231 is stored. For this reason, the open process of the log file 231 on the shared disk 230 results in an error.

  Thereafter, the area of the shared disk 230 is mounted from the server device on the active side (server device 210a), and the application 221 that implements the function of the cluster service 220 of the server device 210a is started. The syslogd 211a of the server device 210a is Since the log file 231 of the shared disk 230 is not opened, the message output by the application 221 is not output to the log file 231 of the shared disk 230. In order to log the message output by the application 221 to the log file 231 of the shared disk 230, the active-side server apparatus (server apparatus 210a) mounts (recognizes) the area of the shared disk 230 and then the syslogd 211a of the server apparatus 210a. Need to restart.

  Further, in FIG. 7, the log service 211 is executed from the syslogd 211a on the server device 210a to the log file 231 of the shared disk 230, and the cluster service 220 is changed from the server on the active side (server device 210a) to the standby side. When switched to the server (server apparatus 210b), the server apparatus 210a tries to unmount (detach) the shared disk 230 area. However, since the file open for the log file 231a of the shared disk 230 continues from the syslogd 211a on the server device 210a side, this unmount process results in an error, and the cluster service 220 switching process fails. Alternatively, the syslogd 211a itself on the active side (server apparatus 210a) is stopped by clustering software (software for cooperative configuration).

  Note that the technique of Patent Document 1 described above can record a temporarily stored log message in a log file on a shared disk in a cluster system using an active server and a spare server using a shared disk. The problem that an error occurs in the process of opening the log file on the shared disk is still not solved.

  The technique of the above-mentioned patent document 2 can monitor the output data of the application program by temporarily storing the log data output from the application in the named pipe, but this named pipe technique is used. Even so, it cannot be applied to cluster service switching processing in a cluster system. Further, even if the technique of Patent Document 1 and the technique of Patent Document 2 described above are combined, the timing of file open / file close of the log file on the shared disk is not taken into consideration. The problem that sometimes causes errors still remains.

  The present invention has been made in view of the above-described circumstances, and is an Active- that prevents an error from occurring when switching between the active server and the spare server by appropriately recording data in a log file on a shared disk. It is an object of the present invention to provide a stand-by type cluster system, a method for recording data in a log file on a shared disk in the cluster system, and a program.

  The present invention has been made to solve the above problem, and is a cluster system in which an active server and a standby server operate using a shared disk, and log messages of an application are stored on a local disk. Storage means for temporarily storing in a named pipe, recording means for recording the log message temporarily stored in the named pipe in a log file on the shared disk, and the recording means storing the log message in the log file There is provided a cluster system comprising timing setting means for setting timing for recording in a log file.

  The present invention also relates to a method for recording data in a log file on a shared disk in a cluster system composed of an active server and a standby server, and temporarily logs an application log message to a named pipe on the local disk. Set in the first step, the second step for setting the timing to record the log message temporarily stored in the named pipe to the log file, and the second step And a third step of recording the log message temporarily stored in the named pipe to a log file on the shared disk based on timing. A method for recording data in a file is provided.

  The present invention also relates to a program in which a cluster system composed of an active server and a standby server records data in a log file on a shared disk, and the computer logs the application log message to the name on the local disk. Storage means for temporarily storing in the attached pipe, recording means for recording the log message temporarily stored in the named pipe in the log file on the shared disk, and the recording means storing the log message in the log file Provided is a program for functioning as timing setting means for setting a recording timing.

  According to the present invention, in an Active-Standby type cluster system, a log message processed via syslogd for log message transfer is stored in a log file on a shared disk by a program that reads a named pipe (FIFO). Thus, the message processed by the syslogd can be properly recorded in the file on the shared disk. As a result, even if a failure occurs on the active server (active active server) and a failover occurs to switch to the standby server (standby spare server), the standby server can transfer to the shared disk. You can continue to add logs to the same log file.

It is a block diagram which shows a structure when the Active-Standby type cluster system which concerns on embodiment of this invention designates a log file on a shared disk. It is a flowchart which shows the flow of operation | movement of the starting process in the cluster system shown in FIG. 2 is a flowchart showing a flow of processing for outputting a log message to a log file on a shared disk after the cluster system shown in FIG. 1 is activated. 3 is a flowchart showing a flow of stop processing in the cluster system shown in FIG. 1. It is a flowchart which shows the flow of a process when failing over the cluster service of the cluster system shown in FIG. It is a block diagram which shows the structure of a general Active-Standby type cluster system. It is a block diagram which shows a structure when a general Active-Standby type cluster system designates a log file on a shared disk.

  The cluster system according to the embodiment of the present invention is an Active-Standby type cluster system that uses a shared disk, and a message that is processed via a syslogd having a transfer function by a program that reads a named pipe (FIFO). It is characterized by saving to a log file on a shared disk. That is, in the Active-Standby type cluster system of this embodiment, when a message processed by the syslogd is recorded in a log file on the shared disk, a named pipe on the local disk (as a log file output destination on the syslogd side ( FIFO). As a result, since the log file opened by syslogd exists on the local disk, the file opening process is performed even when the shared disk is not mounted. In addition, the shared disk can be unmounted even when syslogd is activated.

  Further, a message output from the syslogd to a named pipe (FIFO) is read by a program (log application), and this message is recorded in a log file on the shared disk. This program always opens / writes / closes the log file on the shared disk every time a message is output from syslogd to the named pipe, so the log file on the shared disk is always opened. It will not be.

  Hereinafter, a preferred embodiment of a recording method for a log file on a shared disk in a cluster system according to the present invention will be described in detail with reference to the drawings.

<Embodiment>
FIG. 1 is a block diagram showing a configuration when an Active-Standby type cluster system according to an embodiment of the present invention specifies a log file on a shared disk. First, the configuration of the cluster system shown in FIG. 1 will be described. This cluster system includes server devices 10a and 10b (hereinafter, when the server devices 10a and 10b are collectively referred to as the server device 10) and a shared disk 30. In the present embodiment, the server device 10a functions as the active side (active system), and the server device 10b functions as the standby side (standby system). Therefore, the provision of the cluster service 20 is performed by the server device 10a on the active side.

  The shared disk 30 includes data necessary for providing the cluster service 20, and is used only from the server device 10 on the active side, that is, the server device 10a. The syslogd 11 (syslogd 11a, 11b) having a data transfer function operates on the server device 10, and sends a message to the log file 12 (log files 12a, 12b) and the named pipe 13 (named pipes 13a, 13b). Send.

  The log application 14 (log applications 14a and 14b) of the server apparatus 10 waits for an input from the syslogd 11 to the named pipe 13, and when there is an input from the syslog 11 to the named pipe 13, the log file on the shared disk 30 31 is output.

  Note that the storage means for temporarily storing the log message of the application 21 in the named pipe 13 on the local disk is realized by the syslog 11 of the server device 10, and the log message temporarily stored in the named pipe 13. Is recorded in the log file 31 on the shared disk 30, and the timing setting means for setting the timing at which the recording unit records the log message in the log file 31 is realized by the log application 14 of the server device 10.

  Next, the basic operation flow of the cluster system shown in FIG. 1 will be described with reference to a flowchart. First, the operation flow of the startup process in the cluster system will be described. FIG. 2 is a flowchart showing the operation flow of the startup process in the cluster system shown in FIG.

  First, when the OS (Operating System) of the server apparatus 10 is activated (step S1), the syslogd 11 is activated on the server apparatus 10 (step S2). At this time, the server apparatus 10 opens the named pipe 13 set with the syslogd 11 in the write mode. Next, the log application 14 is activated on the server device 10, reads the named pipe 13, opens a file in a dedicated mode, and waits for input from the syslog 11 (step S3).

Then, when the respective cluster software is activated and communication is performed on the server device 10 (step S4), it is determined which of the server device 10a and the server device 10b is the active side (active system). Is done. That is, the server device 10 determines whether or not the own device is on the active side (step S5). If it is determined that the own device is the active side of the cluster system (Yes in step S5), the server device 10 mounts (recognizes) the area of the shared disk 30 (step S6), and then the cluster service 20 The application 21 that realizes the above function is activated (step S7), and the activation process is completed (step S8). In the example of FIG. 1, the server apparatus 10a performs this operation.
On the other hand, if it is determined in step S5 that the own apparatus is not the active side (No in step S5), the server apparatus 10 performs the activation process operation without starting the application 21 and mounting the area of the shared disk 30. Is completed (step S8). In the example of FIG. 1, the server device 10b executes the operation.

  Next, the flow of processing for outputting a message to the log file on the shared disk after starting the cluster system will be described. FIG. 3 is a flowchart showing a flow of processing for outputting a log message to the log file on the shared disk after the cluster system shown in FIG. 1 is started. Here, the case where the server apparatus 10a is operating as the active side of the cluster system as shown in FIG. 1 will be described.

  In FIG. 3, when the log output operation is started (step S11), first, the application 21 for realizing the cluster service 20 outputs a message to the syslogd 11a on the server device 10a on the active side (step S11). S12). Next, the log application 14a of the server device 10a reads the message via the named pipe 13a, and opens the log file 31 of the shared disk 30 in the write mode (step S13).

  Next, the log application 14a of the server device 10a outputs the read message to the log file 31 of the shared disk 30 (step S14), and when the output of the read message is completed, the log file 31 of the shared disk 30 is closed. (Step S15), the message output to the log file 31 on the shared disk 30 is completed (Step S16).

  Next, the flow of the stop process operation in the cluster system will be described. FIG. 4 is a flowchart showing the flow of the stop process in the cluster system shown in FIG. In FIG. 4, when a stop command is issued to the cluster system and stop processing is started (step S21), the server device 10 determines whether or not the own device is an active server (step S21). S22). When the own device is an active server (Yes in step S22), the stop of the cluster service 20 of the server device 10 is started and the application 21 is stopped (step S23).

Then, after the application 21 is stopped, the area of the shared disk 30 is unmounted (separated) (step S24). Furthermore, after the area of the shared disk 30 is unmounted, the server apparatus 10 stops the cluster software as an OS stop process (step S25).
Further, the server device 10 stops the log application 14 (step S26), then stops the syslogd 11 (step S27), and completes the stop processing of the server device 10 (step S28).

  If it is determined in step S22 that the own apparatus is not the server on the active side (No in step S22), the server apparatus 10 passes step S23 and step S24 and executes the process of step S25.

  Next, an operation flow when failover (switching processing) of the cluster service of the cluster system to an alternative server will be described. FIG. 5 is a flowchart showing a processing flow when failing over the cluster service of the cluster system shown in FIG.

  First, when a cluster service switching command is issued to the cluster system and failover is started (step S31), the server apparatus 10 determines whether or not the own apparatus is an active server (step S32). ). When the own device is an active server (Yes in step S32), the stop of the cluster service 20 of the own device is started and the application 21 is stopped (step S33).

  After the application 21 is stopped, the area of the shared disk 30 is unmounted (separated) (step S34), and the failover (switching process) of the cluster service is completed (step S38).

  On the other hand, in step S32, when the own device is not the server on the active side (No in step S32), the server device 10 completes the switching process of the partner device (or the server device 10a when the own device is the server device 10). And waits for the completion of the switching process (step S35).

  When the switching from the partner device to the own device is completed (Yes in step S35), the standby server device mounts the area of the shared disk 30 (step S36), and then starts the application 21 as the cluster service 20 ( In step S37), the failover (switching process) of the cluster service is completed (step S38).

  The embodiments of the cluster system according to the present invention have been described in detail with reference to the drawings. However, the specific configuration of the present invention is not limited to these embodiments, and the scope of the present invention is not deviated from. Even if there are changes in the design, etc., they are included in the present invention.

  The operation of the server device 10 described above is stored in a computer-readable recording medium in the form of a program, and the above-described processes are performed by the computer reading and executing this program. Here, the computer-readable recording medium means a magnetic disk, a magneto-optical disk, a CD-ROM, a DVD-ROM, a semiconductor memory, or the like. Alternatively, the computer program may be distributed to the computer via a communication line, and the computer that has received the distribution may execute the program.

  Further, the above program may be for realizing a part of the functions described above. Furthermore, what can implement | achieve the function mentioned above in combination with the program already recorded on the computer system, what is called a difference file (difference program) may be sufficient.

1, 1a, 1b server device 11, 11a, 11b syslogd (storage means)
12, 12a, 12b, 31 log (log file)
13, 13a, 13b Named pipes 14, 14a, 14b Log application (recording means, timing setting means)
20 Cluster service 21 Application 30 Shared disk

Claims (9)

A cluster system in which the active server and standby server operate using a shared disk,
A storage means to temporarily store application log messages in a named pipe on the local disk;
Recording means for recording the log message temporarily stored in the named pipe to a log file on the shared disk;
Timing setting means for setting a timing at which the recording means records a log message in the log file;
A cluster system characterized by comprising:   The cluster system according to claim 1, wherein the recording unit records the log message in the log file by a program that reads the named pipe.   3. The recording unit according to claim 1, wherein the recording unit specifies the named pipe existing on a local disk of the active server or the standby server, and records the log message in the log file. The cluster system described in 1.   4. The timing setting unit performs open / write / close processing of the log file at a timing when the log message is output from the storage unit to the named pipe. The cluster system described in A method of recording data in a log file on a shared disk in a cluster system composed of an active server and a standby server,
A first step of temporarily saving an application log message to a named pipe on the local disk;
A second step of setting a timing for recording the log message temporarily stored in the named pipe to the log file;
A third step of recording the log message temporarily stored in the named pipe in a log file on the shared disk based on the timing set in the second step;
A method for recording data in a log file on a shared disk in a cluster system, comprising:   6. The method for recording data in a log file on a shared disk in a cluster system according to claim 5, wherein the log message is recorded in the log file by a program that reads the named pipe.   7. The log message is recorded in the log file by designating the named pipe existing on a local disk of the active server or the standby server. A method of recording data in a log file on a shared disk in the described cluster system.   8. The cluster system according to claim 5, wherein the log file is opened / written / closed at a timing when the log message of the application is stored in the named pipe. Data recording method to log file on shared disk. A cluster system consisting of an active server and a standby server is a program that records data in a log file on a shared disk,
Computer
A storage means to temporarily store application log messages in a named pipe on the local disk,
Recording means for recording the log message temporarily stored in the named pipe to a log file on the shared disk;
Timing setting means for setting a timing at which the recording means records a log message in the log file;
Program to function as.

Images