KR20040051042A - Balk process method for high availability system - Google Patents

Abstract

PURPOSE: A method for processing a failure of a high availability system is provided to precisely detect/process the failure of a node by reexamining a state of the node generating the failure to all nodes if the failure occurs, and sharing an examining result. CONSTITUTION: If the failure is detected from the specified node, all nodes respectively reexamine a failure state. All nodes transmit each state information to all other nodes through a private network. After collecting the state information of other nodes, each node checks the failure state of the specified node by integrating the state information.

Description

Fault handling in high availability systems {BALK PROCESS METHOD FOR HIGH AVAILABILITY SYSTEM}

The present invention relates to a failure handling method of a high availability system. In particular, when a failure occurs in a specific node, the failure of a specific node is precisely detected by re-checking the state of the specific node in which the failure occurs and sharing the test result. To troubleshoot a high availability system.

When a node included in the high availability configuration fails during operation, the high availability system obtains the status information of the failed node and takes over the operation that was operated on the failed node by the remaining nodes included in the high availability system configuration. It is operated continuously to minimize downtime of the service.

In the conventional high availability system, when a specific node fails, all nodes included in the high availability configuration retest the failure state to update the status information of the corresponding node and respond to the failure.

Nodes included in the high availability configuration, the method of re-checking the failure of a specific node, a method through a private network (public network), a method through a public network, a shared file system channel (file system) channel).

Here, whether or not the method through the shared file system channel can be used depends on a takeover configuration of a service performed by a node.

At this time, the node failure state test result by retesting, failure of the high availability daemon of the node, failure of the private network (private network), failure of the private network and public network (public network) This is classified as a node down failure.

1 is a schematic diagram of a high availability system consisting of four nodes and two services, wherein the first performing node of service 1 is node 1, the turnover order is node 1 → node 2 → node 4, and the first performing node of service 2 is shown. Is node 3, and the turnover order is node 3-> node 4.

When the high availability daemon (Daemon 1 ~ Daemon 4) is executed in all nodes to perform the high availability system, the first performing node of the services among the nodes performs the service, and after all the services are performed, all nodes have their own Check the status of other nodes and services running on other nodes.

At this time, since the first performing node of the service 1 is node 1, the node 1 performs service 1, and the first performing node of the service 2 is node 3, so that the node 3 performs service 2 of all nodes. High availability daemons exchange state information with each other over a private network and check the status of different nodes.

If a failure occurs in a specific node, the remaining nodes cannot receive the status information of the corresponding node. If the status information of the specific node is not received in this way, the failure of the corresponding node is detected and the failure is re-inspected.

The retesting sequence is a sequence of inspection through a private network → inspection through a public network → inspection through a shared file system channel. same.

First, if a response from the node is checked through a private network, it is determined that a failure occurs in the high availability daemon of the node. If there is no response, '2' of FIG. 2 is performed.

Next, if a response is received from the corresponding node through the public network, it is determined that a failure occurs in the private network of the node, and if there is no response, '3' of FIG. 3 is performed.

In the node that can be checked through the shared file system channel according to the service takeover order, if there is a response from the node by checking through the shared file system channel, the node's private It is determined that a failure occurs in a private network and a public network, and when there is no response, a node down failure is determined.

On the other hand, a node that cannot be inspected through the shared file system channel according to the service handover sequence is determined as a node down failure as shown in FIG.

For example, if node 1's private network and public network fail, node 2 and node 4 are included in the node taking over service 1, and thus the shared file system channel. Since inspection is possible through the (file system channel), it is determined through the shared file system channel to determine if the private network and the public network of Node 1 have failed.

At this time, since node 3 is not included in the node taking over service 1, inspection is impossible through the shared file system channel, so it is determined that the node down failure has occurred.

However, since all nodes included in the above-described high availability configuration have their respective failure states re-checked to update the status information of the corresponding nodes and respond to the failures, each node may determine the failure differently for the same failure. In this case, there is a problem in that the node state information of the high availability system is inconsistent and the administrator misidentifies the state of the system.

The present invention has been made to solve the above problems, and when a failure occurs in a specific node, by re-checking the state of the specific node where the failure occurs in all nodes and sharing the test result, the failure of the specific node is accurately detected. The purpose of the present invention is to provide a failure handling method of a high availability system.

1 is a block diagram of a high availability system consisting of four nodes and two services.

FIG. 2 is a diagram showing the inspection via a private network when a daemon failure of Node 1 occurs.

3 is a diagram illustrating the inspection over the public network when a private network failure of node 1 occurs.

Figure 4 shows the inspection over a shared file system channel in the event of a private network and public network failure of node 1 in Figure 1;

5 is a flowchart illustrating a failure handling method of the high availability system of the present invention.

FIG. 6 is a schematic diagram showing an individual inspection operation of each node when a private network and a public network failure of node 1 occur in FIG. 5; FIG.

FIG. 7 shows an operation of transmitting a retest result of each node in FIG. 5; FIG.

The present invention for achieving the above object, the first process of re-checking the failure state of each node, if a failure of a particular node is detected; A second process in which all nodes transmit respective state information to all other nodes through a private network; After each node collects state information of other nodes, the node performs a third process of checking the failure state of a specific node by synthesizing the state information.

Hereinafter, with reference to the accompanying drawings, the operation and effects of the failure handling method of the high availability system according to the present invention will be described in detail.

FIG. 5 is a flowchart illustrating an operation of a failure handling method of a high availability system of the present invention, as shown in the first step of determining whether a failure occurs in a specific node; As a result of the determination, when a failure occurs in a specific node, a second process of rechecking a failure state of all nodes respectively; A third step of all nodes transmitting respective state information to all other nodes through a private network; After collecting the state information of the other nodes, the fourth step of identifying the failure state of a specific node by combining the state information, this operation of the present invention will be described.

First, an operation of a failure handling method of a high availability system according to the present invention will be described as an operation when a failure occurs in a private network and a public network of a node 1, which is a conventional example. do.

First, when a specific node fails, for example, after a high availability system is performed, when a private network and a public network of node 1 fail, all nodes each have their own nodes as shown in FIG. The state of node 1 is determined.

Here, in FIG. 5, since the node 2 and the node 4 are inspected through the private network, there is no response from the node 1 (1), and the inspection is performed through the public network. Since there is no response at node 1 (2), the check is performed through a shared file system channel, and as a result of the test, there is a response at node 1 (3), so node 1's status is not associated with the private network. Determine if the public network has failed.

In the case of Node 3, since there is no response from Node 1 as a result of the test through the private network (4), the test is performed through the public network, and the response is returned from Node 1 as the test result. And it is impossible to perform a check over a shared file system channel (5), so it is determined to be a node down.

Then, as shown in FIG. 7, all nodes transmit a result of determining the failure state of Node 1 to another node, and Node 2 and Node 4 fail in the private network and the public network. The node 3 transmits the result determined by the node down.

Then, after each node collects the results received by other nodes and collects the information, the node 1 status is equally determined by all nodes in the private network and the public network. network failure).

In this case, the criterion for determining the aggregated information may be to determine a failure state of a specific node by assigning a priority to a case where a private network and a public network failure occur, rather than a case where a node down failure occurs.

The specific embodiments or examples made in the detailed description of the present invention are intended to clarify the technical contents of the present invention to the extent that they should not be construed as limited to these specific examples and should not be construed as consultations. Various changes can be made within the scope of.

As described in detail above, the present invention maintains consistency of high availability state information by re-checking the state of a specific node in which the failure occurs and sharing the test result when all nodes fail when the high availability system operates. It is effective.

Claims (2)

If a failure of a specific node is detected, a first step of re-checking each failure state of all nodes; A second process in which all nodes transmit respective state information to all other nodes through a private network; After each node collects the status information of the other nodes, the third step of identifying the failure status of a specific node by combining the status information, the failure handling method of a high availability system. The method of claim 1, wherein the third process comprises: When the private network and the public network failure occurs, giving a higher priority than the node down failure occurs, and confirming the failure status of a specific node comprising the step of handling failure of a high availability system.