Search

CN-121979686-A - Cloud host management method, device and medium

CN121979686ACN 121979686 ACN121979686 ACN 121979686ACN-121979686-A

Abstract

The embodiment of the invention provides a cloud host management method, a cloud host management device and a cloud host management medium, wherein the cloud host management method comprises the steps of creating a task by utilizing a time trigger mechanism and an event trigger mechanism, task information of the task comprises task resources, determining a first target domain agent corresponding to the task by utilizing a task planning agent according to the task information, detecting the task resources corresponding to the task by utilizing the first target domain agent, determining a first target tool identifier by utilizing the first target domain agent according to a mapping relation between the domain agent and the tool identifier, acquiring a target tool instance corresponding to the first target tool identifier from a registry, detecting the task resources corresponding to the task, and generating a suggestion notification according to a task processing result. According to the cloud host management method and device, timeliness, integrity and accuracy of cloud host management can be effectively improved under the condition that manual operation cost and processing time delay are reduced, and adaptability to different management requirements can be enhanced.

Inventors

  • HU ZHIQIANG
  • Ai Tianxiang

Assignees

  • 天翼云科技有限公司

Dates

Publication Date
20260505
Application Date
20260403

Claims (20)

  1. 1. A method for managing a cloud host, the method comprising: Creating a task by using a time trigger mechanism and an event trigger mechanism, wherein task information of the task comprises task resources, wherein the task resources comprise cloud hosts and associated resources of the cloud hosts in all resource pools under a user account and target resources corresponding to preset events; determining a first target domain intelligent agent corresponding to the task according to the task information by utilizing a task planning intelligent agent; The method comprises the steps of detecting task resources corresponding to a task by utilizing a first target domain intelligent agent to obtain a corresponding task processing result, wherein the first target domain intelligent agent determines a first target tool identifier according to a mapping relation between the domain intelligent agent and the tool identifier, acquires a target tool instance corresponding to the first target tool identifier from a registry, and detects the task resources corresponding to the task; generating a suggestion notification according to the task processing result; and processing the advice notification.
  2. 2. The method according to claim 1, wherein the method further comprises: storing, with the hierarchical configuration file, a plurality of agent nodes organized in a tree structure, each agent node comprising an associated agent node or an associated tool identifier; Storing tool identification and tool execution information of a tool by using a tool configuration file; generating a tool instance according to the tool execution information in the tool configuration file by using the registry, and registering the mapping relation between the tool identifier and the corresponding tool instance; The first target domain agent determines a first target tool identifier according to a mapping relationship between the domain agent and the tool identifier, including: Determining a target agent node corresponding to the first target domain agent from the hierarchical configuration file; And if the target intelligent agent node comprises the associated intelligent agent node, determining the first target tool identifier according to the intelligent agent node associated with the target intelligent agent node.
  3. 3. The method according to claim 2, wherein the method further comprises: and adding tool identification and tool execution information of the newly added tool in the tool configuration file, and adding the tool identification of the newly added tool aiming at the agent node corresponding to the newly added tool in the hierarchical configuration file.
  4. 4. A method according to any one of claims 1 to 3, wherein creating tasks using a time trigger mechanism and an event trigger mechanism comprises: Under the condition that the triggering time is reached, a first task is created, wherein the first task is used for detecting cloud hosts and associated resources of the cloud hosts in all resource pools under a user account; and under the condition that the preset event is captured, creating a second task, wherein the second task is used for detecting a target resource corresponding to the preset event.
  5. 5. A method according to any one of claims 1 to 3, wherein determining, by using a task planning agent, a first target domain agent corresponding to the task according to the task information includes: and inputting a first prompt word to the task planning agent, wherein the first prompt word comprises task information and information of a first target domain agent, so that the task planning agent determines the first target domain agent corresponding to the task according to the matching information between the task information and the information of the domain agent.
  6. 6. The method of any one of claims 1 to 3, wherein the domain agents include a computing agent, a network agent, and a storage agent; the computing agent corresponds to task resources of a computing type, the network agent corresponds to task resources of a network type, and the storage agent corresponds to task resources of a storage type.
  7. 7. A method according to any one of claims 1 to 3, wherein the detecting, with the first target domain agent, a task resource corresponding to the task includes: Determining the residual time length of the cloud host and the associated resources of the cloud host in all the resource pools under the user account by using the first target domain intelligent agent, and marking the cloud host and the associated resources of the cloud host with the residual time length smaller than a preset threshold value, or Determining load states of cloud hosts and associated resources of the cloud hosts in all resource pools under a user account by using the first target domain intelligent agent, and marking the resources with the load states meeting set conditions, or Identifying the cloud host and the idle state of the related resources of the cloud host in all the resource pools under the user account by utilizing the first target domain intelligent agent, and marking the resources with the idle state conforming to the idle condition, or And detecting the abnormality of the target resource corresponding to the preset event by using the first target domain intelligent agent.
  8. 8. A method according to any one of claims 1 to 3, wherein generating a advice notification according to the task processing result comprises: inputting a second prompt word to the generating agent to obtain a suggestion notification output by the generating agent; The second prompt word comprises second system prompt information, second historical memory information and third tool information, the second system prompt information comprises second role information, second rule information and the task processing result, the second historical memory information comprises user historical behavior data, and the third tool information comprises a tool list for calling.
  9. 9. The method according to any one of claims 1 to 3, wherein the type of the advice notification includes a manual operation type or an execution type; The processing of the advice notification comprises: if the advice notification is of the manual operation type, the advice notification containing the operation guidance information is sent to the target user, or And if the advice notification is of an execution type, sending execution inquiry information to a target user, and calling a tool in the advice notification according to the authorization information of the user for the execution inquiry information to execute the operation corresponding to the advice notification.
  10. 10. A method according to any one of claims 1 to 3, wherein the method further comprises: receiving a user request; Determining user intention corresponding to the task by utilizing a task planning agent, and generating an execution plan corresponding to the user intention, wherein the execution plan comprises a subtask sequence, subtask information corresponding to each subtask comprises a second target domain agent for executing the subtask, a second target tool identifier, input parameters and execution conditions; And calling a tool corresponding to the second target tool identifier by using the second target domain intelligent agent so as to execute the subtask sequence.
  11. 11. The method of claim 10, wherein the utilizing a mission planning agent to determine a user intent corresponding to the mission and generating an execution plan corresponding to the user intent comprises: inputting a third prompting word into the task planning agent to obtain an execution plan output by the task planning agent; The third prompt word comprises third system prompt information, third historical memory information and third tool information, wherein the third system prompt information comprises third color information, third rule information and user intention, the third historical memory information comprises user historical behavior data, the third tool information comprises a tool list for calling and a hierarchical configuration file, the hierarchical configuration file is used for storing a plurality of agent nodes organized in a tree structure, and each agent node comprises an associated agent node or an associated tool identifier.
  12. 12. The method of claim 10, wherein the execution plan comprises a fixed plan or a dynamic plan, wherein the fixed plan characterizes an execution plan that has been completely determined by the sequence of sub-tasks at the time of execution plan generation, and wherein the dynamic plan characterizes an execution plan of a subsequent sub-task that is adjusted during execution based on the execution results of the preceding sub-task.
  13. 13. The method of claim 10, wherein the execution plan is a dynamic plan, and wherein invoking the second target tool to identify a corresponding tool using the second target domain agent to execute the sequence of sub-tasks comprises: Sequentially executing the 1 st to nth sub-tasks in the dynamic plan by using the second target field agent to obtain an execution result of the n sub-tasks; a fifth prompt word containing the execution results of n subtasks is sent to a task planning agent so as to obtain a new subtask output by the task planning agent; and executing the new subtasks by using the second target domain intelligent agent.
  14. 14. The method of claim 10, wherein invoking the second target tool to identify the corresponding tool with the second target domain agent to perform the sequence of subtasks comprises: Inputting a fourth prompt word to the second target field agent to obtain an execution result of the ith subtask; The fourth prompting word comprises fourth system prompting information, fourth flow prompting information, fourth historical memory information and fourth tool information; The fourth system prompt information comprises fourth color information, fourth rule information and subtask information of the ith subtask; The fourth flow prompt information comprises an execution result of a subtask positioned before the ith subtask; the fourth historic memory information comprises user historic behavior data; The fourth tool information includes a tool list for calling.
  15. 15. The method according to claim 10, wherein the method further comprises: presetting key nodes for each subtask; Storing a dialogue identifier corresponding to the subtask and an execution context corresponding to the subtask into a database under the condition that the execution of the subtask reaches the key node; under the condition that a session restarting request is received, searching in a database according to a session identifier corresponding to the session restarting request to obtain an execution context corresponding to the session restarting request; Displaying the abstract and the recovery options of the break point operation to a target user; And restoring the execution of the subtasks from the key nodes based on the execution context data according to a confirmation instruction of a user for a restoration option.
  16. 16. The method of claim 15, wherein the execution context comprises input parameters collected by the subtask and sources thereof, historical tool call records of the subtask, operations to be confirmed of the subtask, current key node locations of the subtask.
  17. 17. The method of claim 15, wherein the break point operation summary comprises a summary of key nodes for which the subtask has been completed, a description of current key nodes for the subtask, and contents of operations to be validated for a standby decision.
  18. 18. A management device of a cloud host, the device comprising: The task information of the task comprises task resources, wherein the task resources comprise cloud hosts and associated resources of the cloud hosts in all resource pools under a user account and target resources corresponding to preset events; The first task planning module is used for utilizing task planning agents and determining first target field agents corresponding to the tasks according to the task information; The task processing module is used for detecting task resources corresponding to the task by utilizing the first target domain intelligent agent to obtain a corresponding task processing result, wherein the first target domain intelligent agent determines a first target tool identifier according to a mapping relation between the domain intelligent agent and the tool identifier, acquires a target tool instance corresponding to the first target tool identifier from a registry, and detects the task resources corresponding to the task; the suggestion generation module is used for generating a suggestion notification according to the task processing result; and the suggestion processing module is used for processing the suggestion notification.
  19. 19. An electronic device comprising a processor and a non-transitory memory having instructions thereon, wherein the instructions, when executed by the processor, cause the processor to perform the method of any of claims 1-17.
  20. 20. A non-transitory computer readable storage medium storing instructions that cause a processor to perform the method of any one of claims 1-17.

Description

Cloud host management method, device and medium Technical Field The embodiment of the invention relates to the technical field of cloud computing, in particular to a management method, a device and a medium of a cloud host. Background In the field of cloud computing, the management of a cloud host refers to the management and control of the whole process from creation, starting, running and configuration change to final deletion of the cloud host, and is used for realizing the dynamic scheduling and on-demand allocation of computing resources and storage resources related to the cloud host, improving the utilization rate of the whole resources, reducing the idle cost, supporting the failover or fast reconstruction when hardware faults or system anomalies occur so as to shorten the service downtime, and realizing the fine control of the cost by combining the real-time load monitoring and dynamic adjustment of instance specifications. In the current cloud host management method, a cloud host is generally managed according to an operation request submitted by a user. For example, by calling an operation parameter such as a resource pool ID (Identity), an available area name, an instance ID, etc. where an application programming interface is transferred into a target cloud host, operations such as creating, starting, suspending, restarting, reloading, deleting, etc. are performed on the target cloud host. According to the current cloud host management method, a user needs to check the resource state to judge the state of due renewal, low load idling and the like, and the judgment is time-consuming and labor-consuming and is easy to miss. Under the condition of missing judgment, low-load idling can cause resource waste, and service interruption can be caused when the service is not renewed due to expiration. Disclosure of Invention The embodiment of the invention provides a management method, a device and a medium for a cloud host, which can effectively improve the timeliness, the integrity and the accuracy of cloud host management and can enhance the adaptability to different management requirements under the condition of reducing the manual operation cost and the processing time delay. In a first aspect, an embodiment of the present invention discloses a method for managing a cloud host, where the method includes: Creating a task by using a time trigger mechanism and an event trigger mechanism, wherein task information of the task comprises task resources, wherein the task resources comprise cloud hosts and associated resources of the cloud hosts in all resource pools under a user account and target resources corresponding to preset events; determining a first target domain intelligent agent corresponding to the task according to the task information by utilizing a task planning intelligent agent; The method comprises the steps of detecting task resources corresponding to a task by utilizing a first target domain intelligent agent to obtain a corresponding task processing result, wherein the first target domain intelligent agent determines a first target tool identifier according to a mapping relation between the domain intelligent agent and the tool identifier, acquires a target tool instance corresponding to the first target tool identifier from a registry, and detects the task resources corresponding to the task; generating a suggestion notification according to the task processing result; and processing the advice notification. In a second aspect, an embodiment of the present invention discloses a management apparatus for a cloud host, where the apparatus includes: The task information of the task comprises task resources, wherein the task resources comprise cloud hosts and associated resources of the cloud hosts in all resource pools under a user account and target resources corresponding to preset events; The first task planning module is used for utilizing task planning agents and determining first target field agents corresponding to the tasks according to the task information; The task processing module is used for detecting task resources corresponding to the task by utilizing the first target domain intelligent agent to obtain a corresponding task processing result, wherein the first target domain intelligent agent determines a first target tool identifier according to a mapping relation between the domain intelligent agent and the tool identifier, acquires a target tool instance corresponding to the first target tool identifier from a registry, and detects the task resources corresponding to the task; the suggestion generation module is used for generating a suggestion notification according to the task processing result; and the suggestion processing module is used for processing the suggestion notification. In a third aspect, an embodiment of the present invention discloses an electronic device comprising a processor and a non-transitory memory having instructions thereon, wherein the instruction