JP2003345620A - Process monitoring method for multi-node cluster system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To always grasp movements of a cluster group (transition of states) and to automatically change monitor settings according to them. <P>SOLUTION: Servers 2 to 4 to be monitored including a standby server 4 are equipped with monitor processes 6, 7, and 14 for detecting abnormality of the cluster group and check processes 15, 16, and 17 for acquiring information B for checking whether the cluster group move or stop. A monitor setting process 11 of a monitor server 1 is informed of the check information B by the servers 2, 3, and 4 to be monitored and compares the check information B with a monitor setting table 13 in a memory 12 by the servers 2, 3, and 4 to be monitored to judge that the cluster group is switched between the servers 2 and 3 to be monitored and the standby server 4 when a discrepancy is found, thereby changing monitor settings for monitoring the servers to be monitored through a monitor process 5. <P>COPYRIGHT: (C)2004,JPO

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a method for monitoring a server process in a multi-node cluster system.

[0002]

2. Description of the Related Art Conventionally, as a method for monitoring a server process, a method described in JP-A-2001-117789 has been known. This is provided with a program monitoring condition setting unit, which can be set as a program monitoring condition by an external input device such as a graphical user interface or a definition file, and a program name to be monitored can be set with an arbitrary character string first, and a program can be set secondly. The names of all or some of the constituent processes can be set by the name of the executable file or the name of the command line. Thirdly, the lower limit and upper limit of the number of processes that are determined to be operating normally for each process name specified by the second setting Can be set as a threshold, and a wide range of process monitoring by the number of processes is realized.

[0003]

In recent years, a stand-by server common to a plurality of systems has been prepared from the viewpoints of improving reliability and reducing costs, and in which system a process has been switched to a stand-by system. However, there is a multi-node cluster system that switches to the same standby server. However, with the current process monitoring method, it is difficult to constantly monitor the process on the standby server in the multi-node cluster system.

[0004] This will be described below with reference to FIG.

In FIG. 1, a three-node cluster system including one monitoring server 1, two monitored servers 2 and 3, and one standby server 4 will be described as an example.

The monitoring server 1 has a monitoring server-side monitoring process (hereinafter, simply referred to as a monitoring process) 5 which is always started.
To monitor. Further, the monitoring target server 2 starts a cluster group 8 that is composed of one or more processes 10 and forms a package (program), and monitors the cluster group 8 with the process 10 as a monitoring target. (Hereafter simply referred to as the monitoring process)
6 is provided. The monitoring result of the monitoring process 6 is sent to the monitoring process 5 of the monitoring server 1 as monitoring target process information A. Similarly, the monitored server 3
A monitoring target side monitoring process (hereinafter, simply referred to as a monitoring process) 7 that activates a cluster group 9 that is configured by one or more processes 10 and forms a package (program) and monitors the cluster group 9 with the process 10 as a monitoring target. It has. The monitoring result of the monitoring process 7 is sent to the monitoring process 5 of the monitoring server 1 as monitoring target process information A.

In the monitoring process 5, for each of the cluster groups 8 and 9 of the monitored servers 2 and 3, monitoring settings such as a process monitoring item and a range of the number of processes for determining that the cluster group is normal are set. The status of the cluster groups 8 and 9 is monitored by comparing the monitored process information A from the monitored servers 2 and 3 with the corresponding monitoring settings.

Here, the activated cluster group 8
When a failure occurs in the monitored server 2 due to an abnormality in the monitored server 2 and the monitored server 2 cannot execute the cluster group 8, switching to execute the cluster group 8 on the standby server 4 is performed, that is, The standby server is switched. In such a case, it is conceivable that the monitoring server 1 monitors the standby server 4. For this purpose, the monitoring process 5 of the monitoring server 1 is performed.
First, the monitoring setting for the cluster group 8 in the standby server 4 must be performed.

However, in order to do this,
It is conceivable to provide a monitoring process in the standby server 4 and to notify the monitoring process 5 of the monitoring server 1 of the monitoring result. However, when the system is in a normal state, the standby server 4 is running. In such a case, the monitoring process 5 of the monitoring server 1 determines that the monitoring process 5 of the monitoring server 1 is based on the monitoring target process information for the status from the monitoring process of the standby server and the monitoring setting for the standby server. It is determined that the server 4 has a failure.

[0010] Therefore, if the standby server is switched from the monitored server 2 to the cluster group 8 in this state, the monitoring process 5 in the monitoring server 1 monitors the standby server 4 for the monitoring corresponding to the cluster group 8. The setting must be changed (updated). However, in the system having the configuration shown in the figure, it is not possible to automatically change the monitoring setting to the corresponding setting when the standby server is switched. For this reason, at present, the standby server 4
Is not monitored.

Therefore, after the standby server is switched, the state of the cluster group 8 running on the standby server 4 cannot be monitored, and even if an abnormality occurs, there is no way to know it. Was.

SUMMARY OF THE INVENTION An object of the present invention is to solve such a problem, to constantly grasp the movement (change of state) of a cluster group, and to automatically change a monitoring setting in accordance with the movement. It is an object of the present invention to provide a process monitoring method adapted to cope with the switching of a standby system.

[0013]

In order to achieve the above object, the present invention provides a monitoring server, a monitored server whose cluster group is monitored by the monitoring server, and a standby server common to the monitored server. In the process monitoring method of a multi-node cluster system, the monitored server and the standby server respectively check the process management table of the operating system at regular time intervals by a check process that is always running, and start and stop the cluster group. The confirmation result indicating the stop is notified to the monitoring setting change process on the monitoring server.
Based on the result of the notification, the switch of the cluster group between the monitored server and the standby server is recognized, and based on this recognition in the monitoring setting change process, the monitoring that the cluster group is running It monitors processes of the target server and the standby server.

The monitoring server is provided with a monitoring setting table indicating the correspondence between the activated cluster group, the monitored server, and the standby server, and the monitoring settings corresponding to the activated cluster group. The monitoring setting change process compares the confirmation result notified from the check process with the monitoring setting table for each of the monitored server and the standby server, and compares the cluster between the monitored server and the standby server. The switching of the group is recognized, and the monitoring setting table is changed together with the recognition of the switching.

Further, in accordance with the recognition of the switching of the cluster group by the monitoring setting change process, a process of changing the monitoring setting for the monitored server or the standby server in which the cluster group has been started or stopped is performed.

The monitoring setting is provided in the monitoring server in advance for each cluster group and in accordance with the combination of the cluster groups, and the monitoring target server started by the cluster group or one or more cluster groups are started. The standby server is monitored by selecting and setting the corresponding monitoring setting.

[0017]

Embodiments of the present invention will be specifically described below with reference to the drawings. 1 to 4 are system diagrams showing an embodiment of a process monitoring method for a multi-node cluster system according to the present invention, wherein 11 is a monitoring setting change process, 12 is a memory, 13 is a monitoring setting table, 14
Is a monitoring process, 15 to 17 are check processes, and portions corresponding to those in FIG. 7 are denoted by the same reference numerals. Here, as in the case of FIG. 7, a case of a three-node cluster system will be described as an example.

FIG. 1 shows a normal process monitoring mode.

In FIG. 1, each of the monitored servers 2 and 3 is provided with a check process 15 and 16, and the standby server 4 is provided with a monitor process 14 and a check process 17. It also works as well. In the monitoring server 1, a monitoring setting change process 11 and a memory 12 are provided.
The monitoring setting table 13 for 3 and 4 is stored so as to be writable and readable.

The monitoring processes 6 and 7 of the monitored servers 2 and 3 started by the cluster groups 8 and 9 acquire the monitored process information A of the corresponding cluster groups 8 and 9 and monitor the monitored process information A of the monitoring server 1. Notify process 5.
In the monitoring process 5, monitoring settings for each of the cluster groups 8 and 9 (respectively referred to as monitoring settings a and b) and monitoring settings for a combination of the cluster groups 8 and 9 (referred to as monitoring settings c) are provided. Monitoring server 2
Is a monitoring setting a for the cluster group 8
However, for the monitored server 3, the cluster group 9
Of the cluster groups 8 and 9 by comparing the notified monitoring target process information A with the corresponding monitoring setting for each of the monitoring target servers 2 and 3. are doing. The monitoring process 5 determines, for example, that the number of processes in the cluster group 8 is out of the specified range or that the process has terminated abnormally based on the monitoring target process information A and the monitoring setting a from the monitoring target server 2. When the monitoring target server 2 is recognized, the monitoring target server 2 determines that there is an abnormality, and executes a set process (for example, a process of sounding a patrol lamp or notifying an administrator of a warning message).

The above operation is almost the same as that of the conventional system, however, in this embodiment, the monitored servers 2, 3,
4 and the monitoring server 1 in the above-described configuration, the original standby server 4 is also included as a monitoring target server, and the movement (state transition) of the cluster groups 8 and 9 is constantly monitored. The monitoring process 5 can automatically perform the correct monitoring setting for the corresponding monitoring target servers 2, 3, and 4.
In order to make this possible, check processes 15 and 16 are added to the monitored servers 2 and 3, and a monitoring process 14 and a check process 17 are provided in the standby server 4 to configure a monitored server. In the server 1, a monitoring setting change process 11 and a memory 12 having a monitoring setting table 13 are added.

Here, the monitoring setting change process 11 of the monitoring server 1 and the check processes 15, 16, and 17 of the monitoring target servers 2, 3, and 4 are always running, and the monitoring setting change process 11 and the monitoring target server 2 , 3, and 4 check processes 15, 16, and 17, the processing operations shown in FIG.

That is, the check processes 15, 16, 17
Checks the process management tables of the operating systems of the monitored servers 2, 3, and 4 at regular time intervals (step 100), and uses the check result as the active cluster group information B to change the monitoring setting on the monitoring server 1. The process 11 is notified (step 101). The monitoring setting change process 11 compares the running cluster group information B with the contents of the monitoring setting table 13 in the memory 12 and determines whether or not the cluster groups 8 and 9 are moving (that is, whether the cluster groups 8 and 9 are in a state due to switching of the standby server, etc. Change (transition) is detected).

The monitoring setting table 13 shows information indicating which monitoring target server each of the cluster groups 8 and 9 is running and the monitoring setting for the monitoring target server set by the monitoring process 5 at that time. ing. In the illustrated state of the monitoring setting table 13, the cluster group 8 is activated on the monitored server 2, and the monitoring setting a of the cluster group 8 is selected and set for the monitored server 2 in the monitoring process 5. And that the cluster group 9 is the monitored server 3
Indicates that the monitoring setting b of the cluster group 9 has been selected and set for the monitoring target server 3 in the monitoring process 5, and the monitoring target server (standby server) 4 In this case, since the standby server has not been switched at this time, there is no cluster group to be activated, and therefore, the monitoring process 5 indicates that the monitoring setting has not been selected and set for the monitoring target server 4. I have. Therefore, the monitoring process 5 does not monitor the standby server 4.

The respective monitored servers 2, 3, and 4
When the starting cluster group information B is notified from the check processes 15, 16, and 17 of the monitoring target, the monitoring setting change process 11 executes the starting cluster group information B and the monitoring setting table received for each of the monitored servers 2, 3, and 4. 13 and the information of the corresponding monitored server is compared. In this case, the running cluster group information B from each of the monitored servers 2, 3, and 4 matches the contents of the monitoring setting table 13, so that each time the operation shown in FIG. The processing consisting of steps 200 and 201 is performed.

In the state shown in FIG. 1, if it is assumed that a failure has occurred in the monitored server 2,
Switching of the standby server of cluster group 8
Is switched to the monitored server 4 which is a standby server,
The monitoring target server 4 is started. FIG. 2 shows such a state of the system, and the processing operation in this case will be described with reference to FIG.

In FIG. 2 and FIG. 6, as described above, the monitoring setting change process 11 on the monitoring server 1 receives notification of the active cluster group information B from each of the monitored servers 2, 3, and 4. If there is no standby server switching (step 201),
With the processing operation shown in FIG. 5 at regular time intervals, the operations of steps 200 and 201 are repeated.

However, as described above, the monitored server 2
When the standby server switching occurs only in the standby server, the check process 17 on the monitored server 4 checks the active cluster group 8 in the process management table of the operating system of the monitored server 4 (Step 100 in FIG. 5), Monitoring setting change process 1 on monitoring server 1
1 is notified of the running cluster group information B (step 101 in FIG. 5). At this time, in the monitoring setting table 13 managed on the memory 12 of the monitoring server 1, the information shown in FIG. When the booting cluster group information B is received from the check process 17 of the server 4 (step 200), it is compared with the information for the monitored server 4 in the monitoring setting table 13 shown in FIG. Compare (step 201). At this time, since the running cluster group information B from the check process 17 indicates that only the cluster group 8 has been started on the monitored server 4, the monitoring setting change process 11 indicates this in the monitoring setting table 13. Of the monitored server 4 (that is, information indicating that there is no active cluster group) (step 201), only the cluster group 8 is activated from the state in which there is no active cluster group on the monitored server 4. The monitoring process 5 recognizes that the state has transitioned to the changed state (step 202), and executes the monitoring setting changing process in the monitoring process 5 (step 205). In this change process, the monitoring process 5 selects and sets the monitoring setting a used for monitoring the monitoring target server 2 when the monitoring target server 4 is normal. Thereby, the monitoring process 5 starts monitoring the monitored server 4.

Thereafter, the monitoring setting change process 11
The monitoring setting table 13 on the memory 12 is changed (step 206). In this change processing, only the cluster group 8 is running on the monitored server 4, and at this time, the monitoring setting table 1 is set so that the monitoring setting for the monitored server 4 in the monitoring process 5 is the monitoring setting a.
3 is to be changed.

On the other hand, in the monitored server 2, when the standby server of the cluster group 8 is switched, the check process 15 on the monitored server 2 is also performed by the monitoring server 1.
The running cluster group information B indicating that the running cluster group does not exist is notified to the monitoring setting change process 11 (step 10 in FIG. 5).
1). Upon receiving this notification (step 200), the monitoring setting change process 11 compares the running cluster group information B with the information of the monitoring target server 2 in the monitoring setting table 13 on the memory 12 (step 200). 201), it is recognized that there is no active cluster group (step 202), the monitoring setting a on the monitoring target server 2 in the monitoring process 5 is canceled, and the monitoring of this monitoring server 2 is ended. (Step 20
3). As a result, the monitored server 2 is excluded from the monitoring setting.

The monitoring setting change process 11 uses the starting cluster group information B from the check process 15 of the monitoring target server 2 and the starting cluster group information B from the checking process 17 of the monitoring target server 4. The switching of the standby server of the cluster group 8 of the server 2 can be recognized, and based on this recognition, the monitoring process 5 selects and sets the monitoring setting a for the monitoring target server 4 as described above. be able to.

In addition to the process of terminating the monitoring of the monitored server 2 (step 203) and the selection and setting of the monitoring setting a for the monitored server 4 (step 205), the monitoring setting change process 11 performs the process of changing the monitoring setting table 13. Is executed (step 206). This is to change the monitoring setting table 13 shown in FIG. 1 so as to match the new state of the system. It is assumed that the cluster group 8 is starting up on the monitored server 4 and the monitored server 4 On the other hand, the monitoring setting in the monitoring process 5 is referred to as monitoring setting a, and has the contents as shown in FIG.

When the above process is completed, the monitoring process 5 of the monitoring server 1 monitors the monitored servers 3 and 4 in the same manner as described above, and monitors the monitored servers 2 and 3 respectively.
The check processes 15, 16, and 17 of 4 repeat the operation shown in FIG. 5 at regular time intervals.

In the state shown in FIG. 2 described above,
In some cases, a failure occurs in another monitored server, in this case, the monitored server 3, and the cluster group 9 of the monitored server 3 is switched to the monitored server 4 that is a standby server. FIG. 3 shows such a state, and the processing operation for this will be described below with reference to FIGS.

In FIG. 3 and FIG. 6, as described above, the monitoring setting change process 11 on the monitoring server 1 receives notification of the active cluster group information B from each of the monitored servers 2, 3, and 4. If there is no standby server switching (step 201),
In the state shown in FIG. 2, the operations of steps 200 and 201 are repeated with the processing operation shown in FIG. 5 at regular time intervals.

In this state, when the standby server switching occurs in the monitored server 3 as described above, the check process 17 on the monitored server 4 starts up in the process management table of the operating system of the monitored server 4. The middle cluster groups 8 and 9 are confirmed (step 100 in FIG. 5), the combination check process is executed, and the monitoring setting change process 11 on the monitoring server 1 is performed.
Is notified of the activated cluster group information B indicating the combination check processing (step 101 in FIG. 5).

This combination check process is called when the number of activated cluster groups becomes two or more. Each of the activated cluster groups is checked and set in the monitoring server 1 in advance. This is a process for selecting a monitoring setting corresponding to the combination of cluster groups (in this case, the above-described monitoring setting c) from the monitoring settings that have been set.

At this time, in the monitoring setting table 13 managed on the memory 12 of the monitoring server 1,
Although the information shown in FIG. 2 is maintained as it is without being changed, the monitoring setting change process 11 receives the starting cluster group information B from the check process 17 of the monitoring setting server 4 (step 200). This is compared with the information for the monitored server 4 in the monitoring setting table 13 shown in FIG. 2 managed on the memory 12 (step 201). At this time, since the running cluster group information B from the check process 17 indicates that the cluster groups 8 and 9 have been started on the monitored server 4, the monitoring setting change process 11 indicates this to the monitoring setting table 13. From the state in which only the cluster group 8 is running on the monitored server 4 by comparing with the information on the monitored server 4 in step (i.e., information indicating that only the cluster group 8 is running) (step 201). Recognizing that the cluster groups 8 and 9 have transitioned to the activated state (step 202), the monitoring process 5 executes a process of changing the monitoring settings accordingly. In this change process, a combination of two or more cluster groups is confirmed (step 204: in this case, it is confirmed that the combination is a combination of two cluster groups 8 and 9). In response to this, the above-mentioned monitoring setting c corresponding to such a combination is selected and set (step 205). As a result, the monitoring process 5 continues monitoring the monitoring target server 4.

Thereafter, the monitoring setting change process 11
The monitoring setting table 13 on the memory 12 is changed (step 203). In this change processing, the monitoring setting table 1 is set so that the cluster groups 8 and 9 are running on the monitored server 4 and the monitoring setting for the monitored server 4 in the monitoring process 5 is the monitoring setting c at this time.
3 is to be changed.

On the other hand, in the monitored server 3, when the standby server of the cluster group 9 is switched, the check process 16 on the monitored server 3 is also performed by the monitoring server 1.
The running cluster group information B indicating that the running cluster group 9 does not exist is notified to the monitoring setting change process 11 (step 10 in FIG. 5).
1). Upon receiving this notification (step 200), the monitoring setting change process 11 on the monitoring server 1 compares the running cluster group information B with the information of the monitoring target server 3 in the monitoring setting table 13 on the memory 12. As a result (step 201), it is recognized that the active cluster group does not exist (step 20).
2) The monitoring setting b of the monitoring target server 3 in the monitoring process 5 is released, and the monitoring of the monitoring server 3 is terminated (step 203). As a result, the monitored server 3 is also excluded from the monitoring setting.

In this case as well, the monitoring setting change process 11 includes the starting cluster group information B from the check process 17 of the monitored server 4 and the starting cluster group information B from the check process 16 of the monitored server 3. The cluster group 9 of the monitored server 3
Can be recognized, and based on this recognition, the monitoring process 5 can select and set the monitoring setting c for the monitoring target server 4 as described above. .

In addition to the process of terminating the monitoring of the monitored server 3 (step 203) and the selection and setting of the monitoring setting c for the monitored server 4 (step 205), the monitoring setting change process 11 performs the process of changing the monitoring setting table 13. Is executed (step 206). This changes the monitoring setting table 13 shown in FIG. 2 so as to match the new state of the system. The cluster group 9 is also activated on the monitored server 4 and the monitored server 4 On the other hand, the monitoring setting in the monitoring process 5 is referred to as monitoring setting c, and has the contents shown in FIG.

Next, a description will be given of the operation in the case where the monitored server in which a failure has occurred is recovered, and the cluster group which has been activated in the standby server returns (switches back) to the original monitored server.

In some cases, the system enters the state shown in FIG. 3, and thereafter the cluster group 8 running on the monitored server 4 which is the standby server returns to the original monitored server 2 (this will be described below). Standby server switchback). FIG. 4 shows such a state, and the processing operation for this will be described below with reference to FIGS.

4 and 6, as described above, the monitoring setting change process 11 on the monitoring server 1 receives notification of the active cluster group information B from each of the monitored servers 2, 3, and 4. If the state transition of the cluster groups 8 and 9 does not occur (step 201) and the processing shown in FIG. 5 is performed at regular time intervals in the state shown in FIG. Operation is repeated.

In this state, as described above, when the standby server returns from the monitored server 4 where the cluster group 8 is switched to the monitored server 2, the check process 17 on the monitored server 4 performs this process. Only the running cluster group 9 is checked in the process management table of the operating system of the monitored server 4 (step 100 in FIG. 5), and the running cluster group information B indicating this to the monitoring setting change process 11 on the monitoring server 1 (Step 101 in FIG. 5). At this time, in the monitoring setting table 13 managed on the memory 12 of the monitoring server 1, the information shown in FIG. Check process 1 of server 4
7 (step 200), it compares this with the information for the monitored server 4 in the monitoring setting table 13 shown in FIG. 3 managed in the memory 12 (step 201). ). At this time, since the running cluster group information B from the check process 17 indicates that only the cluster group 9 has been started on the monitored server 4, the monitoring setting change process 11 indicates this in the monitoring setting table 13. From the state in which the cluster groups 8 and 9 are running on the monitored server 4 by comparing with the information on the monitored server 4 (that is, information indicating that only the cluster group 9 is running) (step 201). Recognizing that only the cluster group 9 has transitioned to the activated state (step 202), the monitoring process 5 executes a process of changing the monitoring setting in accordance with this. This change process selects and sets the monitoring setting b used for monitoring the monitoring target server 3 in the monitoring process 5 when the monitoring target server 3 is normal. Thus, the monitoring process 5 continues monitoring the monitoring target server 4 with the monitoring setting b (step 205).

After that, the monitoring setting change process 11
The monitoring setting table 13 on the memory 12 is changed (step 206). In this change process, only the cluster group 9 is running on the monitored server 4, and at this time, the monitoring setting table 1 is set so that the monitoring setting for the monitored server 4 in the monitoring process 5 is the monitoring setting b.
3 is to be changed.

On the other hand, in the monitored server 2, when the standby server of the cluster group 8 returns, the check process 15 on the monitored server 2 also performs the monitoring process 1.
The monitoring setting change process 11 is notified of the starting cluster group information B indicating that the starting cluster group 8 is present (step 101 in FIG. 5).
Upon receiving this notification (step 200), the monitoring setting change process 11 on the monitoring server 1 executes the startup cluster group information B and the information of the monitoring target server 2 in the monitoring setting table 13 shown in FIG. (Step 201), it is recognized that the activated cluster group 8 is in a state of being present (step 2).
02), a monitoring setting change process is executed by the monitoring process 5 (step 205). This change process selects and sets a monitoring setting a used by the monitoring process 5 to monitor the monitoring target server 2 in a normal state. Thereby, the monitoring process 5 starts monitoring the monitoring target server 2.

In this case as well, the monitoring setting change process 11 includes the starting cluster group information B from the check process 17 of the monitored server 4 and the starting cluster group information B from the check process 15 of the monitored server 2. As a result, from the monitored server 4 to the monitored server 2
The switching back of the standby server of the cluster group 8 to the monitoring target server 4 can be recognized, and based on this recognition, the monitoring process 5 transmits the monitoring setting b to the monitoring target server 4 as described above. 2, the monitoring setting a can be selected and set.

In addition to the monitoring start processing of the monitored server 2 (step 205) and the monitoring setting change of the monitored server 4 (step 205), the monitoring setting change process 11 executes the processing of changing the monitoring setting table 13 (step 205). Step 206). This changes the contents of the monitoring setting table 13 shown in FIG. 3 so as to match the new state of the system. The cluster group 9 is the monitored server 4 and the cluster group 8 is the monitored server 2. It is assumed that each of them is running, the monitoring setting of the monitoring process 5 for the monitoring target server 4 at this time is a monitoring setting b, and the monitoring setting of the monitoring target server 2 for the monitoring process 5 is a monitoring setting a. The contents are as shown in FIG.

In the state shown in FIG. 4, the same applies to the case where the cluster group 9 running on the monitored server 4 has the monitored server 3 switched back to the standby server. Since the server 4 switches from a state in which the cluster group 9 is running to a state in which the starting cluster group does not exist, the monitoring of the monitoring process 5 for the monitoring target server 4 is released, and the memory 1
2 has the contents shown in FIG.

As described above, in this embodiment, the monitoring setting change process 11 and the memory 12 on the monitoring server 1 side
With the above monitoring setting table 13, the standby server 4 can also be used as a monitoring target server to constantly monitor the movement (transition) of the cluster group between these monitoring target servers. The monitoring setting corresponding to the server can be accurately and automatically selected and set. Even if the standby server is switched, correct monitoring of the cluster group can be performed.

In the above embodiment, a system in which the number of monitored servers is two, the number of standby servers is one, and the number of monitoring servers is one has been described, but the present invention is not limited to this. Instead, it goes without saying that each of these servers also corresponds to an arbitrary number of systems.
Of course, in this case, not only the monitoring settings for the cluster groups in each monitored server, but also the monitoring settings for all or a part of these cluster groups that are actually feasible are created in advance. This means that the server can be selected for the monitoring process.

In the above embodiment, the monitoring process 5 of the monitoring server 1 recognizes the movement of the cluster group based on the starting cluster group information B from each of the monitored servers 2, 3, and 4 in the monitoring setting change process 11. Based on (Step 202 in FIG. 6), the selection and setting of the monitoring setting of the monitoring target server (Step 205 in FIG. 6) and the monitoring end processing (Step 203 in FIG. 6) are performed. 11 is each monitored server 2, 3, 4
When the movement of the cluster group is recognized based on the startup cluster group information B from the server (step 202 in FIG. 6), first, the monitor setting table 13 in the memory 12 is changed. May be checked to perform the selection / setting of the monitoring setting and the monitoring end processing. In this case, the monitoring process 5 can recognize the movement of the cluster group (switching of the standby server or switching back) based on the monitoring target process information A from at least one of the monitoring target servers. By checking the monitoring setting table 13 changed by the setting changing process 11, it is possible to select and set the monitoring setting and perform the monitoring end processing.

[0055]

As described above, according to the present invention, in a multi-node cluster system, it is also possible to monitor the process of the standby server, and the monitoring target including the standby server according to the movement of the cluster group. The monitoring setting for the server can be correctly and automatically changed, and appropriate process monitoring can always be realized.

Further, by creating a monitoring setting for a combination of a plurality of cluster groups that may be simultaneously activated on the standby server, the combination of the cluster groups can be controlled regardless of the number of servers constituting the cluster system. Also, the monitoring setting and its change can be made, and the versatility and convenience can be improved.

[Brief description of the drawings]

FIG. 1 is a system diagram showing a process monitoring mode in a normal state in an embodiment of a process monitoring method for a multi-node cluster system according to the present invention.

FIG. 2 is a system diagram showing a process monitoring mode after a standby server is switched due to a failure occurring in only one monitored server in the state shown in FIG. 1;

FIG. 3 is a system diagram showing a process monitoring mode after a standby server switchover due to a standby server switchover occurring in still another monitored server in the state shown in FIG. 2;

FIG. 4 is a system diagram showing a process monitoring mode when the cluster group of the standby server switches back to the standby server in the state shown in FIG. 3;

FIG. 5 is a flowchart showing a specific example of a processing operation of a check process 7 in the monitoring target server in FIGS. 1 to 4;

FIG. 6 is a flowchart showing a specific example of a monitoring setting change processing operation of the embodiment shown in FIGS. 1 to 4;

FIG. 7 is a system diagram showing an example of a conventional process monitoring method of a multi-node cluster system.

[Explanation of symbols]

1 monitoring server 2-3 Monitoring target server 4 Monitored (standby) server 5 Monitoring process on the monitoring server 6,7 Monitoring process on the monitored server 8,9 cluster group 10 Processes to be monitored 11 Monitoring setting change process 12 memory 13 Monitoring setting table 14 Monitoring process on the monitored server 15-17 Check process

Claims (5)

[Claims] 1. A process monitoring method for a multi-node cluster system, comprising: a monitoring server; a monitored server whose cluster group is monitored by the monitoring server; and a standby server common to the monitored server. The server and the standby server respectively check the process management table of the operating system at regular time intervals by a check process that is constantly running, and monitor the monitoring results on the monitoring server to indicate the start and stop of the cluster group. Notifying the setting change process, the monitor setting change process receives the notification of the confirmation result, recognizes the switching of the cluster group between the monitored server and the standby server, and Based on the recognition in the above, the monitored server in which the cluster group is running and the standby server A multi-node cluster system process monitoring method characterized by monitoring a process with a server. 2. The monitoring server according to claim 1, wherein the monitoring server stores a correspondence between the active cluster group, the monitored server, and the standby server, and a monitoring setting corresponding to the active cluster group. A monitoring setting table is provided, and for each of the monitored server and the standby server, the monitoring result is compared with the confirmation result notified from the checking process by the monitoring setting change process. A multi-node cluster system process for recognizing switching of a cluster group between the monitored server and the standby server, and changing the monitoring setting table with the recognition of the switching. Monitoring method. 3. The monitoring setting change for the monitoring target server or the standby server in which a cluster group is started or stopped in accordance with the recognition of the switching of the cluster group by the monitoring setting change process according to claim 1 or 2. A multi-node cluster system process monitoring method, characterized by performing processing. 4. The monitoring server according to claim 3, wherein the monitoring setting is provided in the monitoring server in advance for each cluster group and in accordance with a combination of cluster groups, and A multi-node cluster system process monitoring method, wherein the standby server started by the cluster group is monitored by selecting and setting the corresponding monitoring setting. 5. A multi-node cluster system comprising a monitoring server, a monitored server whose cluster group is monitored by the monitoring server, and a standby server common to the monitored server. A multi-node cluster system, characterized by executing the multi-node cluster system process monitoring method according to any one of the above.