JP2500598B2 - Automatic failure recovery method for network management system - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a failure automatic restoration method for a network management system which is composed of a manager and a plurality of agents and manages management objects such as modems and communication programs.

[0002]

2. Description of the Related Art A conventional network composed of a manager and a plurality of agents, the manager issuing management operations to the agents, and the agents managing management objects such as modems and communication programs in accordance with the management operations from the manager. In the management system, when a failure occurs in the management target, the following processing is performed.

When a failure occurs in a management target such as a modem, a failure occurrence notification is sent to an agent managing the failure.

Upon receiving the failure notification, the agent notifies the manager that the failure has occurred, and performs recovery processing on the failed management target according to the recovery method specified for each management target. After completion of the recovery processing, a failure recovery result notification indicating whether the recovery has succeeded or failed is given to the manager.

Upon receiving the notification of the failure occurrence and the failure recovery result notification, the manager manages them as a failure history and displays the notification contents on the display device to notify the operator of the failure occurrence and the recovery result ( For example, Sugimoto
Osamu, Masao Shimogo "Total Network Management System" NE
C Technical Report, Vol. 40 No.11, Volume 219, p
80-p84, November 10, 1987).

[0006]

As described above, in the related art, since the recovery processing is performed only for the management object to which the failure occurrence notification has been issued, there are the following problems.

That is, when a failure occurs in a management target under a certain agent, the influence may reach a management target under another agent. For example, when a failure occurs in a modem under the control of an agent, the communication program under the control of another agent receiving data via the modem does not receive the data for a certain period of time. It may be determined that a failure has occurred, the failure notification may be sent to the agent, and the agent may end abnormally.

In such a case, if both the modem and the communication program notify the agent of a failure occurrence, the modem and the communication program are respectively restored.

However, a failure may occur in which the failure notification is not sent from the modem to the agent. In such a case, in the above-mentioned conventional technique, only the communication program is restored (for example, the restoration processing is performed). ,Reboot)
The recovery process is not performed for the modem. Therefore, even if the communication program is restarted, the failure of the modem still remains, and the communication program ends abnormally again.

In such a case, the conventional failure automatic recovery method for the network management system cannot automatically recover the failure, and the operator cannot notify the failure occurrence notification and the failure recovery result notification displayed on the display device by the manager. It was necessary to guess the management target (modem) in which a failure occurred based on the contents, and instruct the agent managing the management target that the failure had occurred in the recovery method.

Therefore, in the conventional failure automatic recovery method for a network management system, a failure occurs in a management target under another agent due to a failure of a management target under a certain agent, and a failure occurs in the management target under the certain agent. If the agent is not notified of a failure to be managed, there is a problem that the recovery process takes time and burdens the operator.

An object of the present invention is that a failure occurs in a management target under another agent due to a failure of a management target under a certain agent, and the failure of the management target under the certain agent is notified to the agent. An object of the present invention is to provide a failure automatic recovery method for a network management system that can automatically perform recovery processing in a short time and without burdening an operator even when the failure is not caused.

[0013]

In order to achieve the above object, the present invention comprises a manager and a plurality of agents, each of which manages a management target under its own agent according to an instruction from the manager. In the system, a database that stores recovery methods for multiple agents for each failure information is provided , and each agent receives a failure occurrence notification from the management object under it, thereby performing recovery processing for the management object where the failure occurred. By performing a recovery process failure, by performing a failure occurrence notification including failure information to the manager, the manager receives a failure occurrence notification including failure information from the agent,
The database is searched to find the same failure information as the failure information included in the failure notification, and the recovery method for a plurality of agents stored corresponding to the found failure information is acquired and restored. The agent that has acquired the method is instructed to perform the recovery process according to the acquired recovery method.

[0014]

In the database, for each piece of failure information, a recovery method for an agent that manages a management target in which a failure indicated by the failure information is managed, and management that may cause a failure that causes the failure The recovery method for a plurality of agents is stored for each failure information such that the recovery method for the agent managing the target is stored.

Each agent receives a failure occurrence notification from a subordinate management object to perform a recovery process on the failed management object, and when the recovery process fails, the agent includes failure information for the manager. Notify failure occurrence.

Upon receiving the failure notification, the manager
The database is searched to find the same failure information as the failure information included in the failure notification, and further, a recovery method for a plurality of agents stored corresponding to the found failure information is acquired. Then, it instructs each agent that has acquired the recovery method to perform recovery processing according to the acquired recovery method.

[0017]

Embodiments of the present invention will now be described in detail with reference to the drawings.

FIG. 1 is a block diagram of an embodiment of the present invention, in which a network management system 1 including a manager 2, a database 3 and a plurality of agents 4-1 to 4-i, and a network management system 1 manages the network. Management targets 5-11 to 5-im.

The manager 2 is a subordinate agent 4-1.
It has a function of instructing management operations to 4-i and receiving various notifications from agents 4-1 to 4-i.

The agents 4-1 to 4-i manage each management target 5-11 according to the management operation instructed by the manager 2.
It has a function of actually managing ~ 5-im.

The managed objects 5-11 to 5-im are a modem, a communication control device, a communication program, and the like which compose the network managed by the network management system 1.

The database 3 stores information used when recovering from a failure.

FIG. 2 is a diagram showing an example of the contents of the database 3. Fault information 11-1 to 11-n, agent identifiers 12-11 to 12-nq, and recovery method 13-111.
~ 13-nqu and priorities 14-111 to 14-nqu
And the retry count limit values 15-111 to 15-nqu are stored.

The fault information 11-1 to 11-n indicate the fault occurrence position and contents.

Each fault information 11-1, ..., 11-n
The agent identifier 12-of the agent that gives an instruction when the failure indicated by the
11-12-1p, ..., 12-n1-12-nq are stored.

Further, each agent identifier 12-11,
..., recovery method at the time of failure 13-1 corresponding to 12-1p
11-13-11r, ..., 13-1p1-13-1ps
And each recovery method 13-111 to 13-11r, ..., 13
-1p1 to 13-1ps priority 14-111 to 14-
11r, ..., 14-1p1 to 14-1ps and recovery methods 13-111 to 13-11r, ..., 13-1p1 to 1
Limit value 15-1 for the number of retries for recovery processing at 3-1 ps
11-15-11r, ..., 15-1p1-15-1ps
And are stored.

3 and 4 show the network management system 1
Is a flow chart showing a processing example of the above, and the operation of the present embodiment will be described below with reference to each drawing.

Now, for example, assume that a failure occurs in the managed object 5-i1 under the control of the agent 4-i. Management target 5
When a failure occurs, -i1 notifies the agent 4-i of the failure.

The agent 4-i receiving this notification is
The recovery process for the management target 5-i1 is tried according to a predetermined recovery method for the management target 5-i1.

If the restoration is successful, the agent 4-i finishes the restoration process for the managed object 5-i1. If the recovery fails, the agent 4-i
Judges that the failure cannot be recovered by itself, and notifies the manager 2 of the failure occurrence as shown in FIG. 3 (step S21). This notification includes failure information indicating what kind of failure has occurred in the managed object 5-i1.

When the manager 2 receives the fault occurrence notification from the agent 4-i, it searches the database 3 and
The same fault information as the fault information included in the fault occurrence notification is searched from the fault information 11-1 to 11-n (step S22). Now, for example, it is assumed that the failure information 11-1 has been found.

Upon finding the fault information 11-1, the manager 2 obtains the agent identifiers 12-11, ..., 12-1p stored corresponding to the fault information 11-1.
(Step S23), further, the restoration method 13 having the highest priority 14-111, ..., 14-1p1 for each agent identifier 12-11, ..., 12-1p.
-111, ..., 13-1p1 and retry count limit value 15
-111, ..., 15-1p1 is acquired (step S2
4).

After that, the manager 2 changes to the agent 4
For the agents indicated by the agent identifiers 12-11 to 12-1p acquired in step S23 among -1 to 4-i, the recovery processing by the recovery method 13-111, ..., 13-1p1 acquired in step S24 is performed. While instructing to try, the retry count limit values 15-111, ..., 15-1p1 acquired in step S24 are passed (step S25).

Of the agents 4-1 to 4-i, the agents indicated by the agent identifiers 12-11 to 12-1p each receive the fault recovery instruction from the agent 2 as shown in FIG. 4 (step S4).
1) First, "0" is set to the number of trials (step S4
2) It is determined whether the number of trials is larger than the retry count limit value sent from the manager 2 (step S
43).

Then, the determination result of step S43 is NO.
In the case of, agent identifiers 12-11 to 12-1p
The agent corresponding to (1) tries the recovery process according to the instruction from the manager 2 for the management target under it and increases the number of trials by 1 (step S44).

When the failure recovery is successful (YES in step S45), a failure recovery result notification indicating successful recovery is sent to the manager 2 as shown in FIG.
(Step S47).

If failure recovery fails (NO in step S45), the process returns to step S43.

After that, when the above-mentioned processing is repeated and the judgment result in step S43 becomes YES, as shown in FIG. 6, a failure recovery result notification indicating failure of recovery is sent to the manager 2 (step S46).

The agent 2 executes step S25 in FIG.
When the processing of step 1 is performed, the agent identifiers 12-11 to 1
Based on the failure recovery result notification from the agent corresponding to 2-1p, the agent identifiers 12-11 to 12-
It is determined whether or not the recovery is successful in all the agents corresponding to 1p (step S26). That is, the manager 2 determines that the agent identifier 12-11
Only when all the failure recovery result notifications from 12-1p are received and all of them indicate successful recovery, it is judged that the failure recovery is successful.

When it is determined that the failure recovery has succeeded (YES in step S26), the manager 2 recognizes that the recovery processing has succeeded and ends the recovery processing.

When it is determined that the failure recovery has failed
In (NO in step S26), the manager 2 determines whether or not there is an agent that has exhausted all the recovery methods (step S27).

Then, the determination result of step S27 is NO.
In this case, the manager 2 determines that the agent whose failure recovery has failed (in this case, the agent identifiers 12-11 to 1
The recovery method corresponding to the failure recovery in all the agents corresponding to 2-1p), the recovery method with the highest priority among unretrieved ones among the retry count limit values, 13-112, ..., 13-1p2 are acquired from the database 3 (step S28), and then the process of step S25 is performed.

Thereafter, the same processing as described above is performed,
If there is an agent that has exhausted all the recovery methods (YES in step S27), the manager 2
Recognizes that the recovery process has failed and ends the process.

Since this embodiment operates as described above,
Conventionally, when recovery processing could not be performed automatically, that is, when a management target under a certain agent fails, a management target under another agent also fails.
Even in the case where the agent under the control of a certain agent does not notify the agent of the occurrence of a failure, the recovery processing can be automatically performed as described below.

Now, for example, the management target 5-11 is a modem,
The managed object 5-i1 is a communication program for receiving data via the modem 5-11.

When a failure occurs in the modem 5-11, the communication program 5-i1 cannot receive data for a certain period of time or more, so it judges that the own program 5-i1 has a failure, and the agent 4-i1. A fault occurrence notification is sent to and the process ends abnormally.

At this time, if the fault occurrence notification is sent from the modem 5-11 to the agent 4-1, the fault recovery for the modem 5-11 and the communication program 5-i1 is performed by the agents 4-1 and 4-i. The process is executed.

However, for some reason, the modem 5-
If the failure occurrence notification is not sent from the agent 11 to the agent 4-1, the cause of the failure of the communication program 5-i1 is considered to be even if the agent 4-i restores the communication program 5-i1. Modem 5-11
Since the failure of the above is not recovered, the communication program 5-i1
Cannot be restored.

In such a case, the communication program 5-i
As shown in FIG. 3, the agent 4-i managing 1 manages failure notification to the manager 2 (step S
21).

When the manager 2 receives from the agent 4-i a failure occurrence notification including failure information indicating that the communication program 5-i1 has terminated abnormally, the manager 2 searches the database 3 and searches for the failure information 11-1 to 11-11. Among -n, the one having the same content as the failure information is selected (step S2).
2). Now, for example, it is assumed that the failure information 11-2 shown in FIG. 7 is selected.

After selecting the fault information 11-2, the manager 2 further acquires the agent identifiers 12-21 and 12-22 corresponding to the fault information 11-2 (step S23). In the case of this example, the agent identifier 12-2
1, 12-22 are agents 4-1 and 4-i, respectively.
It is the agent identifier of.

After that, the manager 2 restores the agent identifiers 12-21 and 12-22, the restoration methods 13-211 and 13-221 corresponding to the highest priority 14-211, 14-221, and the retry count limitation. Value 1
5-211 and 15-221 (step S2
4).

Thereafter, the manager 2 restores the agents 4-1 and 4-i indicated by the agent identifiers 12-21 and 12-22 to the recovery method 13-21, respectively.
1, 13-221 is instructed to try the recovery process. That is, in this example, the agent 4-1 is instructed to "switch to the spare circuit in the modem 5-11", and the agent 4-i is restarted "the communication program 5-i1." Command ”(step S2
5).

According to this instruction, the agents 4-1 and
The 4-i performs recovery processing for the modem 5-11 and the communication program 5-i1 and notifies the manager 2 of a failure recovery result (FIG. 4).

The manager 2 has agents 4-1 and 4-
Based on the notification of the failure recovery result from i, the modem 5-11,
It is determined whether or not the recovery process for the communication program 5-i1 has succeeded (step S26).

Now, for example, if it is determined that both the recovery processes for the modems 5-11 and 5-i1 have failed,
The manager 2 has the second highest priority restoration method 13-212, 1 for the agents 4-1 and 4-i, respectively.
It is instructed to try the recovery process by 3-222 (steps S28 and S25). That is, in this example, the agent 4-1 is instructed to "switch the modem 5-11 to the spare modem", and the agent 4-i is reloaded with the communication program 5-i1. , Start up ".

Agents 4-1 and 4 according to this instruction
-I is a modem 5-11 and a communication program 5-
Recovery processing for i1 is performed (FIG. 4), and a failure recovery result notification is sent to the manager 2.

When the manager 2 receives the failure recovery result notification indicating the successful recovery from both the agents 4-1 and 4-i within the predetermined time (step S26), it recognizes that the failure recovery is successful. In all other cases, since all the recovery methods have been tried (YES in step S27), it is recognized that the failure recovery has failed.

[0059]

As described above, the present invention provides, for each failure information, a recovery method for an agent that manages a management target in which a failure indicated by the failure information has occurred,
Stores the recovery method for multiple agents corresponding to multiple failure information, such as the recovery method for the agent that manages the management target that may cause the failure that causes the failure. When a failure notification is sent from an agent, the manager searches the database and follows the recovery method for all the agents that manage the management targets that are considered to be related to the failure. Since a failure occurs in a management target under a certain agent, a failure occurs in a management target under another agent, and Even if the failure is not notified to the agent, it does not burden the operator in a short time. , There is an effect capable of performing recovery processing automatically.

[Brief description of drawings]

FIG. 1 is a block diagram of an embodiment of the present invention.

FIG. 2 is a diagram showing an example of contents of a database 3.

FIG. 3 is a flowchart showing a processing example of the network management system 1.

FIG. 4 is a flowchart showing a processing example of the network management system 1.

FIG. 5 is a diagram showing a sequence when restoration processing is successful.

FIG. 6 is a diagram showing a sequence when recovery processing fails.

FIG. 7 is a diagram showing a part of an example of contents of database 3;

[Explanation of symbols]

 1 ... Network management system 2 ... Manager 3 ... Database 4-1 to 4-i ... Agent 5-11 to 5-im ... Management target 11-1 to 11-n ... Fault information 12-11 to 12-nq ... Agent identifier 13-111 to 13-nqu ... Restoration method 14-111 to 14-nqu ... Priority 15-111 to 15-nqu ... Retry count limit value

Claims (2)

(57) [Claims] 1. A network management system comprising a manager and a plurality of agents, wherein each agent manages a management target under its own agent according to an instruction from the manager, and recovery for a plurality of agents for each failure information By providing a database in which a method is stored , each of the agents receives a failure occurrence notification from a managed object under it, performs a recovery process on the failed managed object, and fails the recovery process,
A failure occurrence notification including failure information is sent to the manager, the manager searches the database by receiving a failure occurrence notification including failure information from the agent, and a failure included in the failure occurrence notification. The same failure information as the information is searched for, and the recovery method for the multiple agents stored corresponding to the found failure information is acquired, and the recovery method for the agent that acquired the recovery method is restored according to the acquired recovery method. An automatic failure recovery method for a network management system, characterized by instructing to perform processing. 2. The database, for each failure information,
A recovery method for an agent that manages the generated managed disorder indicated by the failure information, and a recovery method for an agent for disorders that cause of the disorder is managing managed that can occur Store
Failure automatic recovery process of the network management system of claim 1, wherein the that.