CN111722980B - Data collection systems and methods - Google Patents

Abstract

The embodiment of the invention provides a data acquisition system and a data acquisition method, wherein the system comprises the following steps: the system comprises a plurality of servers, wherein Telegraf nodes are deployed on each server and are used for starting corresponding Telegraf subprocesses to acquire data; the ZooKeeper distributed framework module comprises a plurality of temporary fields registered on the ZooKeeper distributed framework module by the Telegraf node; the Redis database comprises an election event blocking queue for acquiring trigger messages of election events among the plurality of Telegraf nodes. According to the data acquisition system provided by the embodiment of the invention, the Telegraf nodes are respectively deployed on the servers, and when the current Leader node is abnormal, a new Leader node is selected for data acquisition, so that the loss of real-time monitoring data is avoided.

Description

Data collection systems and methods

Technical field

The present invention relates to the field of big data, and in particular to a data collection system and method.

Background technique

Telegraf is a real-time collection tool for performance indicator data related to server-side infrastructure and middleware. The existing technology usually deploys a Telegraf process on a certain server to monitor various infrastructure and middleware distributed on multiple servers.

If the server where the Telegraf process is located experiences problems such as downtime or network failure, causing the Telegraf process to be unable to collect relevant data in real time, the data will be permanently lost because it is not collected in real time, which will affect subsequent data analysis work. cause impact.

Contents of the invention

To address at least one of the above technical problems existing in the prior art, embodiments of the present invention provide a data collection system and method.

In the first aspect, embodiments of the present invention provide a data collection system, including multiple servers, a ZooKeeper distributed framework module and a Redis database, wherein:

The multiple servers are all deployed with Telegraf nodes; the Telegraf nodes included in the data collection system include only one Telegraf node as a Leader node, and the Leader node is used to start a Telegraf sub-process for data collection;

The Telegraf node includes an event listening node and a message acquisition node; the event listening node is used to monitor whether the triggering condition of the Leader node election event is established, and is used to monitor the triggering condition of the Leader node election event when the triggering condition is established. The message that triggers the Leader node election event is sent in the election event blocking queue; wherein the Leader node election event is used to trigger the Telegraf node included in the data collection system to elect a Telegraf node as the Leader node;

The message acquisition node is used to, if the message triggering the Leader node election event is obtained from the election event blocking queue, serve as the elected Leader node and start the corresponding Telegraf sub-process for data collection;

The ZooKeeper distributed framework module includes a temporary field registered by the Telegraf node on the ZooKeeper distributed framework module; wherein the temporary field is used as a trigger condition for the event listening node to monitor the Leader node election event. The basis for whether it is established;

The Redis database includes an election event blocking queue, which is used to store messages that trigger the Leader node election event.

Optionally, the Redis database further includes a distributed lock for determining the Telegraf node that sends the trigger message of the election event to the election event blocking queue.

Optionally, the ZooKeeper distributed framework module also includes a leader field, which is used to store the identification information of the server where the Leader node is located.

In a second aspect, embodiments of the present invention provide a data collection method, applied to the data collection system described in the first aspect, including:

Monitor whether the triggering condition of the Leader node election event is established, and when the triggering condition of the Leader node election event is established, send a message triggering the Leader node election event to the election event blocking queue; wherein, the Leader node election event Used to trigger the Telegraf node included in the data collection system to elect a Telegraf node as the Leader node;

If the message triggering the Leader node election event is obtained from the election event blocking queue, the corresponding Telegraf sub-process will be started to collect data as the selected node that is elected as the Leader node.

Optionally, the triggering conditions for the Leader node election event are specifically:

The temporary field registered by the Leader node on the ZooKeeper distributed framework module does not exist, or the difference between the last reporting time and the current time stored in the temporary field registered by the Leader node on the ZooKeeper distributed framework module exceeds the preset value. threshold.

Optionally, sending a message that triggers the Leader node election event to the election event blocking queue includes:

Multiple Telegraf nodes compete for distributed locks of the Redis database;

The Telegraf node that competes for the distributed lock sends a message that triggers the Leader node election event to the election event blocking queue of the Redis database.

Optionally, after the selected node that is elected as the Leader node starts the corresponding Telegraf sub-process for data collection, it also includes:

If the Telegraf sub-process starts abnormally, the Leader node election event is triggered again.

Optionally, the method also includes:

The ZooKeeper distributed framework module sends an event notification to the Leader node to close the Telegraf sub-process corresponding to the Leader node.

Optionally, the method also includes:

By setting the leader field in the ZooKeeper distributed framework module, one Telegraf node is determined as the Leader node among multiple Telegraf nodes for data collection.

Optionally, by setting the leader field in the ZooKeeper distributed framework module, determine a Telegraf node among multiple Telegraf nodes for data collection, including:

Set the leader field in the ZooKeeper distributed framework module to the identification information of the server where one Telegraf node among multiple Telegraf nodes is located;

The Telegraf node reads the leader field in the ZooKeeper distributed framework module. If the leader field is the identification information of the server where the Telegraf node is located, the Telegraf node starts the corresponding Telegraf sub-process for data collection.

The data collection system provided by the embodiment of the present invention deploys one Telegraf node on each of multiple servers and elects a new Leader node for data collection when the current Leader node is abnormal, thereby avoiding the loss of real-time monitoring data.

Description of the drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the drawings in the following description These are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without exerting creative efforts.

Figure 1 is a schematic structural diagram of a data collection system in an embodiment of the present invention;

Figure 2 is a schematic flow chart of a data collection method in an embodiment of the present invention;

Figure 3 is another schematic flow chart of the data collection method in the embodiment of the present invention;

Figure 4 is another schematic flow chart of the data collection method in the embodiment of the present invention.

Detailed ways

In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments These are some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts fall within the scope of protection of the present invention.

Figure 1 is a schematic structural diagram of a data collection system provided by an embodiment of the present invention. As shown in Figure 1, the system includes:

There are multiple servers 110, and Telegraf nodes 111 are deployed on each server. Among the Telegraf nodes 111 included in the data collection system, there is only one Telegraf node as a leader node. The leader node is used to start the Telegraf sub-process 112 for data collection. ;

The Telegraf node 111 includes an event listening node and a message acquisition node; the event listening node is used to monitor whether the triggering condition of the Leader node election event is established, and is used to monitor whether the triggering condition of the Leader node election event is established, Send a message triggering the Leader node election event to the election event blocking queue; wherein the Leader node election event is used to trigger the Telegraf node 111 included in the data collection system to elect a Telegraf node as the Leader node;

The message acquisition node is used to, if the message triggering the Leader node election event is obtained from the election event blocking queue, start the corresponding Telegraf sub-process 112 as the selected node that is elected as the Leader node. perform data collection;

The ZooKeeper distributed framework module 120 includes a plurality of temporary fields 121 registered by the Telegraf node on the ZooKeeper distributed framework module; wherein, the temporary fields are used by the event listening node to monitor the Leader node election event. The basis for whether the triggering condition is established;

The Redis database 130 includes an election event blocking queue 131, which is used to store messages that trigger the Leader node election event.

Specifically, the present invention is applied in a multi-server environment, and a Telegraf node 111 is deployed on each server 110. That is, there are multiple independent Telegraf nodes 111 in a multi-server environment. One of the important functions of the Telegraf node 111 is to start the corresponding Telegraf sub-process 112 for data collection. More generally, the Telegraf node 111 can be used to manage the life cycle of the corresponding Telegraf sub-process 112, such as starting and stopping the Telegraf sub-process 112.

Telegraf subprocess 112 is a process for monitoring performance indicators related to the operating system server-side infrastructure and middleware. It provides a set of out-of-the-box monitoring data collection plug-ins for many infrastructure and middleware on the operating system server side. Users can configure them as needed. Collect the required plug-ins and adjust the parameters of various plug-ins before they can be put into use in the formal environment. The Telegraf node 111 in the embodiment of the present invention can serve as the parent process of the Telegraf child process 112. In terms of specific implementation, Telegraf node 111 is generally implemented through lightweight scripts such as Bash or Python.

Specifically, the Telegraf node 111 also has the following function: determining the triggering conditions of the election event among the multiple Telegraf nodes 111 and monitoring the triggering message of the election event from the Redis database 130 . The election event between multiple Telegraf nodes 111 refers to: among the multiple Telegraf nodes 111 in the embodiment of the present invention, only one Telegraf node starts the Telegraf sub-process and is responsible for data collection, that is, the Leader node in the embodiment of the present invention. In the Leader When various abnormalities occur in node operation, each Telegraf node 111 needs to elect a new Telegraf node to become the only Leader node, replacing the original Leader node to continue the data collection task.

Furthermore, when various abnormalities occur in the operation of the Leader node, the Telegraf node 111 will determine whether the triggering conditions of the election event are established. When the triggering condition of the election event is established, the trigger message of the election event will appear in the Redis database 130 in the embodiment of the present invention. The Telegraf node 111 will monitor the trigger message of the election event in the Redis database 130 in real time, and can promptly Learn about the trigger of the election event and participate in the election event.

Specifically, the Redis database 130 in the embodiment of the present invention serves as a lightweight message queue. Based on its shared memory working mode, it has better performance when the amount of communication data is small, and is suitable for use as a message queue in the embodiment of the present invention. The target container for election event notification and listening processing. The Redis database 130 specifically includes an election event blocking queue 131. The election event blocking queue 131 is used to obtain the trigger message of the election event between the multiple Telegraf nodes, that is, the Telegraf node sends the trigger message of the election event to the election event blocking queue 131. .

To ensure the uniqueness of the trigger message, only one trigger message of the election event is required when triggering the election event. Therefore, in order to prevent multiple Telegraf nodes from sending election event trigger messages to the election event blocking queue at the same time, the Redis database 130 also contains a distributed lock 132. Multiple Telegraf nodes need to preempt the distributed lock first, and the Telegraf node that preempts the distributed lock Only then can the node send trigger messages to the election event blocking queue, achieving the uniqueness of the trigger messages.

The data collection system in the embodiment of the present invention also includes the ZooKeeper distributed framework module 120. As a program coordination service, the ZooKeeper distributed framework module 120 plays the role of a target container for state synchronization in the system in the embodiment of the present invention, and Its functionality is achieved through the following fields:

Initialization field, used to identify whether the system has completed the initialization process;

The leader field is used to store the identification information of the server where the Leader node is located, such as the host name or IP address of the server;

Temporary fields are multiple temporary fields registered by the Telegraf node on the ZooKeeper distributed framework module. If the Telegraf node is a Leader node, the temporary field registered by the Leader node is used to store the last reporting time of the Leader node; if the Telegraf node is not a Leader node, the temporary field registered by the Telegraf node is used to store Store the health status of the Telegraf node. When the value is true, it means that the node is currently running normally and has the right to be elected as the Leader node. Otherwise, the value is false.

Furthermore, in the embodiment of the present invention, the last reporting time of the Leader node refers to the latest time when the Leader node reports its collected data at a preset frequency during the process of collecting data. The last reporting time is also the last time the Leader node normally collects data. The time it is reported represents the health status of the Leader node.

The data collection system provided by the embodiment of the present invention deploys one Telegraf node on each of multiple servers and elects a new Leader node for data collection when the current Leader node is abnormal, thereby avoiding the loss of real-time monitoring data.

Based on the above embodiments, Figure 2 is a flow chart of the data collection method provided by the embodiment of the present invention. The method is applied to the data collection system provided by the above embodiment. As shown in Figure 2, the method includes:

S201, monitor whether the triggering condition of the Leader node election event is established, and when the triggering condition of the Leader node election event is established, send a message triggering the Leader node election event to the election event blocking queue; wherein, the Leader node The election event is used to trigger the Telegraf node included in the data collection system to elect a Telegraf node as the Leader node;

Specifically, the applied scenario in the embodiment of the present invention is the data collection system of the embodiment of the present invention, and the Telegraf node deployed on a certain server in the system uses the corresponding Telegraf sub-process to perform the data collection task, that is, Leader node.

In order to prevent data collection failure due to problems such as server downtime or network failure, embodiments of the present invention need to elect a new Leader node for data collection. First, each candidate node needs to be controlled to determine whether the triggering conditions for election events between multiple Telegraf nodes are established. Wherein, the candidate node is a Telegraf node other than the Leader node among the plurality of Telegraf nodes. At the same time, the Leader node, like other candidate nodes, actually has the function of determining the triggering conditions of the election event. However, the new Leader node to be elected by the election event is usually generated among the candidate nodes.

Further, the triggering conditions for the election event in the embodiment of the present invention include at least two: the temporary field registered by the Leader node on the ZooKeeper distributed framework module does not exist, or the Leader node is registered on the ZooKeeper distributed framework module. The difference between the last reporting time stored in the temporary field registered on and the current time exceeds the preset threshold. If one of the above two conditions is true, it means that the triggering condition of the election event has been reached and a new Leader node needs to be elected.

Specifically, for the first trigger condition, since each Telegraf node including the Leader node in the embodiment of the present invention will register the corresponding temporary field in the ZooKeeper distributed framework module, if the temporary field corresponding to the Leader node If the status changes from existence to non-existence, according to the event notification mechanism of ZooKeeper, it means that the Leader node no longer exists in the system. Possible reasons include: the server where the Leader node is located is down, the Leader node is disconnected from the ZooKeeper distributed framework module due to network failure. Loss of contact, etc. At this time, the candidate node will receive notification events from the ZooKeeper distributed framework module.

Specifically, for the second trigger condition, the Leader node stores the last reporting time in the temporary field corresponding to the ZooKeeper distributed framework module, and will update the last reporting time at a preset frequency as the data collection task proceeds. Closing time. At the same time, the candidate node will start the corresponding timing check thread to check the last reporting time in the ZooKeeper distributed framework module. If the difference between the last reporting time and the current time exceeds the preset threshold, it means that the Telegraf sub-process controlled by the Leader node reports There are problems such as network congestion and packet loss, which are also situations where election events need to be triggered.

Further, after the candidate node determines that the triggering condition of the election event between multiple Telegraf nodes is established, it needs to send the trigger message of the election event to the election event blocking queue of the Redis database. The election event blocking queue is a message container used to obtain the triggering message of the election event between the multiple Telegraf nodes. That is, in the embodiment of the present invention, the Telegraf node sends the triggering message of the election event to the election event blocking queue. The sender specifically It can be a Leader node or a candidate node.

To ensure the uniqueness of the trigger message, only one trigger message of the election event is required when triggering the election event. Therefore, in order to prevent multiple Telegraf nodes from sending election event trigger messages to the election event blocking queue at the same time, the Redis database also contains distributed locks. Multiple Telegraf nodes need to preempt the distributed locks first, and then the Telegraf node that preempts the distributed locks can Ability to send trigger messages to the election event blocking queue to achieve uniqueness of trigger messages.

S202: If the message triggering the Leader node election event is obtained from the election event blocking queue, as the selected node that is elected as the Leader node, start the corresponding Telegraf sub-process for data collection.

Specifically, after a Telegraf node sends an election event trigger message to the election event blocking queue, according to the exclusivity of the trigger message listening method, only one Telegraf node among multiple Telegraf nodes can listen to the trigger message. That is to say, when one of the Telegraf nodes listens to the trigger message, other nodes can no longer listen to the trigger message. That is to say, only one node obtains votes in the election event and becomes the elected node, that is, New Leader node.

Specifically, the selected node serves as a new Leader node to replace the original Leader node and continue the work of process data collection. The specific steps may be to first start the corresponding Telegraf sub-process by the selected node, and then use the Telegraf sub-process corresponding to the selected node to collect data.

The data collection method provided by the embodiment of the present invention avoids the loss of real-time monitoring data by electing a new Leader node for data collection when the current Leader node is abnormal.

Based on any of the above embodiments, Figure 3 is a flow chart of the data collection method provided by the embodiment of the present invention. As shown in Figure 3, the method is specifically a complete process of the initialization phase, including:

S301, manual random election;

Specifically, when the servers, infrastructure, and middleware in the entire system environment are ready, network operation and maintenance personnel will manually set the leader field in the ZooKeeper distributed framework module to specify the Telegraf node in any server. is the Leader node.

Among them, the leader field is used in the ZooKeeper distributed framework module to store the identification information of the server where the Leader node is located, such as the host name or IP address of the server. Each Telegraf node can learn the information of the Leader node through the information in this field.

S302, start Telegraf node;

Specifically, the initialization field in the ZooKeeper distributed framework module is used to identify whether the system has completed the initialization process. Therefore, at the beginning of the initialization task, you need to manually set the initialization field in the ZooKeeper distributed framework module to false to indicate that the initialization work has not been completed.

At the same time, Telegraf nodes deployed in different servers are generally script programs implemented through lightweight scripts such as Bash or Python. Therefore, to start scripts for multiple Telegraf nodes at the same time during the initialization phase, you can use automated operation and maintenance tools such as Ansible to start Telegraf nodes in batches.

S303, register temporary fields;

Specifically, the ZooKeeper distributed framework module is the target container for state synchronization in the system, and the temporary fields therein are registered by Telegraf nodes on the ZooKeeper distributed framework module, thereby realizing state synchronization of different Telegraf nodes. Therefore, after the Telegraf node is started, the Telegraf node will register with the ZooKeeper distributed framework module with identification information such as the host name or the IP address of the server to obtain the corresponding temporary fields.

S304, obtain Leader information;

Specifically, since the leader field in the ZooKeeper distributed framework module is manually set by network operation and maintenance personnel, each Telegraf node does not yet know whether it is the Leader node set in the initialization task. Therefore, each Telegraf node can read Leader information from ZooKeeper for judgment.

S305, the Leader node starts the Telegraf sub-process;

Specifically, after the Telegraf node reads the Leader information from ZooKeeper, if the Telegraf node finds that it is the Leader node, the Telegraf node starts the corresponding Telegraf sub-process, realizing the Telegraf node specified by the network operation and maintenance personnel in the initialization task. Data collection work.

S306, election event message queue monitoring;

Specifically, when the initialization task specifies the current Leader node for data collection, it also needs to prepare for subsequent election events that may be started. Each Telegraf node needs to know in time when the election event occurs, so each Telegraf node needs to monitor the triggering of the election event. message to achieve.

Specifically, each Telegraf node will start a consumer thread. Specifically, the LPOP/BRPOP mode in Redis can be used to monitor the election event blocking queue in Redis whose current message volume is empty. Once there is an election event trigger message in the election event blocking queue, it will be monitored by each Telegraf node.

S307, initial status check;

Specifically, the Telegraf sub-process is a process directly used for data collection. In order to ensure that there are no abnormalities after the Telegraf sub-process is started, after the Leader node starts the Telegraf sub-process, it is necessary to scan the log file of the sub-process to determine the Telegraf sub-process. Whether it is running normally.

S308, update temporary field status;

Specifically, after the Telegraf node registers with the ZooKeeper distributed framework module to obtain the corresponding temporary field, the status of the temporary field needs to be updated. If the Telegraf node is a Leader node, the temporary field registered by the Leader node is used to store the last reporting time of the Leader node; if the Telegraf node is not a Leader node, the temporary field registered by the Telegraf node is used to store Store the health status of the Telegraf node. When the value is true, it means that the node is currently running normally and has the right to be elected as the Leader node. Otherwise, the value is false.

Therefore, in the step of updating the temporary field status, the Leader node can update the health status of the corresponding temporary node on the ZooKeeper distributed framework module to true after the initial status check is correct; for other nodes, it only needs to determine the election event message queue monitoring If successful, the health status is set to true; if each node cannot listen normally and the Leader node cannot start the Telegraf child process normally, the corresponding health status is set to false.

S309, abnormal alarm detection and execution.

Specifically, after each Telegraf node starts a preset event, such as 30 seconds, the distributed lock will be preempted through Redis, and the Telegraf node that obtains the distributed lock will start an abnormal alarm monitoring thread to monitor each temporary field of the ZooKeeper distributed framework module. The health value is scanned. If false is found, an alarm is triggered, requiring manual intervention for system environment problems. Otherwise, the initialization field of ZooKeeper is set to true, indicating that the initialization of the data collection system is completed.

The data collection method provided by the embodiment of the present invention specifies the Telegraf node responsible for data collection through the initialization task, and implements status synchronization and message monitoring mechanisms among multiple Telegraf nodes, ensuring the normal operation of the data collection task while avoiding Monitor data loss in real time.

Based on any of the above embodiments, Figure 4 is a flow chart of the data collection method provided by the embodiment of the present invention. As shown in Figure 4, the method is specifically a complete process of the election event stage, including:

S401, determine whether an election event is triggered;

In order to prevent data collection from failing due to problems such as downtime or network failure on the server where the original Leader node is located, the embodiment of the present invention needs to elect a new Leader node for data collection in this situation. First, each candidate node needs to be controlled to determine whether the triggering conditions for election events between multiple Telegraf nodes are established. Wherein, the candidate node is a Telegraf node other than the Leader node among the plurality of Telegraf nodes. At the same time, the Leader node, like other candidate nodes, actually has the function of determining the triggering conditions of the election event. However, the new Leader node to be elected by the election event is usually generated among the candidate nodes.

Further, the triggering conditions for the election event in the embodiment of the present invention include at least two: the temporary field registered by the Leader node on the ZooKeeper distributed framework module does not exist, or the Leader node is registered on the ZooKeeper distributed framework module. The difference between the last reporting time stored in the temporary field registered on and the current time exceeds the preset threshold. If one of the above two conditions is true, it means that the triggering condition of the election event has been reached and a new Leader node needs to be elected.

Specifically, for the first trigger condition, since each Telegraf node including the Leader node in the embodiment of the present invention will register the corresponding temporary field in the ZooKeeper distributed framework module, if the temporary field corresponding to the Leader node If the status changes from existence to non-existence, according to the event notification mechanism of ZooKeeper, it means that the Leader node no longer exists in the system. Possible reasons include: the server where the Leader node is located is down, the Leader node is disconnected from the ZooKeeper distributed framework module due to network failure. Loss of contact, etc. At this time, the candidate node will receive notification events from the ZooKeeper distributed framework module.

Specifically, for the second trigger condition, the Leader node stores the last reporting time in the temporary field corresponding to the ZooKeeper distributed framework module, and will update the last reporting time at a preset frequency as the data collection task proceeds. Closing time. At the same time, the candidate node will start the corresponding timing check thread to check the last reporting time in the ZooKeeper distributed framework module. If the difference between the last reporting time and the current time exceeds the preset threshold, it means that the Telegraf sub-process controlled by the Leader node reports There are problems such as network congestion and packet loss, which are also situations where election events need to be triggered.

S402, trigger the election event to the Redis blocking queue;

Specifically, after the candidate node determines that the triggering condition of the election event between multiple Telegraf nodes is established, it needs to send the trigger message of the election event to the election event blocking queue of the Redis database. The election event blocking queue is a message container used to obtain the triggering message of the election event between the multiple Telegraf nodes. That is, in the embodiment of the present invention, the Telegraf node sends the triggering message of the election event to the election event blocking queue. The sender specifically It can be a Leader node or a candidate node.

To ensure the uniqueness of the trigger message, only one trigger message of the election event is required when triggering the election event. Therefore, in order to prevent multiple Telegraf nodes from sending election event trigger messages to the election event blocking queue at the same time, the Redis database also contains distributed locks. Multiple Telegraf nodes need to preempt the distributed locks first, and then the Telegraf node that preempts the distributed locks can Ability to send trigger messages to the election event blocking queue, and release the distributed lock after sending the trigger message, achieving the uniqueness of the trigger message.

S403, nodes that meet the candidate qualifications compete for votes;

Specifically, after a Telegraf node sends an election event trigger message to the election event blocking queue, according to the exclusivity of the trigger message listening method, only one Telegraf node among multiple Telegraf nodes can listen to the trigger message. That is to say, when one of the Telegraf nodes listens to the trigger message, other nodes can no longer listen to the trigger message. In other words, only one node obtains votes in the election event and becomes the elected node, that is, New Leader node.

S404, start the Telegraf sub-process;

Specifically, after the Telegraf node is selected as the new Leader node, the Telegraf node starts the corresponding Telegraf sub-process. When the original Leader node is abnormal, it replaces the Telegraf sub-process in the original Leader node to perform data collection work, ensuring that Data collection continues.

S405, discard the candidacy;

Specifically, after the selected node starts the Telegraf sub-process, it needs to be detected whether the Telegraf sub-process is started successfully. If the Telegraf sub-process starts abnormally, for example, the Telegraf sub-process does not start normally or log scanning finds that reports cannot be received normally after the Telegraf sub-process is started, the Telegraf sub-process and the corresponding consumer thread will be terminated, that is, the candidate will be discarded. Then immediately generate a trigger message for the election event in the election event blocking queue in the Redis database. At the same time, update the health status of the temporary field of the corresponding Telegraf node to false.

S406, synchronize Zookeeper status information;

Specifically, if the selected node detects that the Telegraf sub-process is successfully started, it means that the selected node will officially become a new Leader node and start data collection. At the same time, you also need to update the leader field on the ZooKeeper distributed framework module and update its content to the identification information of the server where the selected node is located. In addition, the selected node, as the new Leader node, also needs to update the last reporting time in the corresponding temporary field.

S407, the original Leader node abdicates smoothly;

Specifically, after the selected node updates the leader field on the ZooKeeper distributed framework module, the original Leader node will receive an event notification from the ZooKeeper distributed framework module. At this time, it will check if the Telegraf sub-process of the original Leader node is still there. If it is running, it will be shut down to achieve smooth switching of the only Telegraf sub-process in the system.

S408, abnormal alarm detection and execution;

Specifically, after step S407, this step is optionally executed to control the selected node to check the number of temporary fields on the ZooKeeper distributed framework module and the number of health values that are true. If one of the values is less than the specified threshold, Indicates that the system is currently in an abnormal state and an alarm is triggered to remind operation and maintenance personnel to intervene in advance.

The data collection method provided by the embodiment of the present invention avoids the loss of real-time monitoring data by electing a new Leader node for data collection when the current Leader node is abnormal.

The device embodiments described above are only illustrative. The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units, that is, they may be located in One location, or it can be distributed across multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Persons of ordinary skill in the art can understand and implement the method without any creative effort.

Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the part of the above technical solution that essentially contributes to the existing technology can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM/RAM, magnetic disk, optical disk, etc., including a number of instructions to cause a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in various embodiments or certain parts of the embodiments.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that it can still be used Modifications are made to the technical solutions described in the foregoing embodiments, or equivalent substitutions are made to some of the technical features; however, these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims (10)

1. A data collection system, characterized by including multiple servers, ZooKeeper distributed framework module and Redis database, wherein: The multiple servers are all deployed with Telegraf nodes; the Telegraf nodes included in the data collection system include only one Telegraf node as a Leader node, and the Leader node is used to start a Telegraf sub-process for data collection; The Telegraf node includes an event listening node and a message acquisition node; the event listening node is used to monitor whether the triggering condition of the Leader node election event is established, and is used to monitor the triggering condition of the Leader node election event when the triggering condition is established. The message that triggers the Leader node election event is sent in the election event blocking queue; wherein the Leader node election event is used to trigger the Telegraf node included in the data collection system to elect a Telegraf node as the Leader node; The message acquisition node is used to, if the message triggering the Leader node election event is obtained from the election event blocking queue, serve as the elected Leader node and start the corresponding Telegraf sub-process for data collection; The ZooKeeper distributed framework module includes a temporary field registered by the Telegraf node on the ZooKeeper distributed framework module; wherein the temporary field is used as a trigger condition for the event listening node to monitor the Leader node election event. The basis for whether it is established; The Redis database includes an election event blocking queue, which is used to store messages that trigger the Leader node election event. 2. The data collection system according to claim 1, wherein the Redis database further includes a distributed lock for determining the Telegraf node that sends the trigger message of the election event to the election event blocking queue. 3. The data collection system according to claim 1, wherein the ZooKeeper distributed framework module further includes a leader field for storing identification information of the server where the Leader node is located. 4. A data collection method, applied to the data collection system according to any one of claims 1 to 3, characterized in that it includes: Monitor whether the triggering condition of the Leader node election event is established, and when the triggering condition of the Leader node election event is established, send a message triggering the Leader node election event to the election event blocking queue; wherein, the Leader node election event Used to trigger the Telegraf node included in the data collection system to elect a Telegraf node as the Leader node; If the message triggering the Leader node election event is obtained from the election event blocking queue, the corresponding Telegraf sub-process will be started to collect data as the selected node that is elected as the Leader node. 5. The data collection method according to claim 4, characterized in that the triggering conditions of the Leader node election event are specifically: The temporary field registered by the Leader node on the ZooKeeper distributed framework module does not exist, or the difference between the last reporting time and the current time stored in the temporary field registered by the Leader node on the ZooKeeper distributed framework module exceeds the preset value. threshold. 6. The data collection method according to claim 4, characterized in that sending a message triggering the Leader node election event to the election event blocking queue includes: Multiple Telegraf nodes compete for distributed locks of the Redis database; The Telegraf node that competes for the distributed lock sends a message that triggers the Leader node election event to the election event blocking queue of the Redis database. 7. The data collection method according to claim 4, characterized in that after the selected node as the Leader node starts the corresponding Telegraf sub-process for data collection, it also includes: If the Telegraf sub-process starts abnormally, the Leader node election event is triggered again. 8. The data collection method according to claim 4, characterized in that the method further includes: The ZooKeeper distributed framework module sends an event notification to the Leader node to close the Telegraf sub-process corresponding to the Leader node. 9. The data collection method according to claim 4, characterized in that the method further includes: By setting the leader field in the ZooKeeper distributed framework module, one Telegraf node is determined as the Leader node among multiple Telegraf nodes for data collection. 10. The data collection method according to claim 9, characterized in that by setting the leader field in the ZooKeeper distributed framework module, determining a Telegraf node among multiple Telegraf nodes for data collection includes: Set the leader field in the ZooKeeper distributed framework module to the identification information of the server where one Telegraf node among multiple Telegraf nodes is located; The Telegraf node reads the leader field in the ZooKeeper distributed framework module. If the leader field is the identification information of the server where the Telegraf node is located, the Telegraf node starts the corresponding Telegraf sub-process for data collection.