Process fault self-healing method, device and device for components in distributed management system

Abstract

The present application discloses a process fault self-healing method, device and device for components in a distributed management system. The method includes: acquiring an application program interface address including a distributed management server in the distributed management system and a component in the distributed management system. Configuration information of the metadata database address of the database; use the API address to obtain the current running status of the component from the distributed management server; use the metadata database address to obtain the metadata from the metadata database; according to the current running status and metadata, the component is fault checked ; When it is detected that there is a faulty component, the application program interface is used to send the process restarting task of the faulty component to the distributed management server. The technical solutions provided by the embodiments of the present application can achieve cross-version compatibility, the self-healing process is visible, the service integration code is non-intrusive, and the process faults of components in the simple and efficient distributed management system are self-healing.

Description

technical field [0001] The present application relates to the technical field of Internet communication, and in particular to a process failure self-healing method, device and equipment for components in a distributed management system. Background technique [0002] With the rapid development of Internet communication technology, some large-scale Internet business systems will adopt distributed cluster management due to business complexity and other reasons. Subsequently, a large number of service management systems for distributed cluster management have also been produced, such as Apache ambari, etc., but with the gradual increase of nodes in the managed single distributed cluster system, when various component failures caused by hardware and software Some common failures such as insufficient memory, network jitter, disk IO overload, etc. cause the process to terminate. Usually, it only needs to restart the process to achieve self-healing. [0003] The existing fault self...

AI Technical Summary

Problems solved by technology

However, there is a problem of version incompatibility between the above two solutions, and when the component process terminates abnormally, the fault self-healing will silently start the process in the background of the distributed management system to recover, resulting in the fault self-healing process can only be done by logging in to the server Checking the logs of the agent node shows that the self-healing process has a black-box nature; and in the two schemes, in the former scheme, the component configuration file needs to be modified and the process restarted, and in the latter scheme, the self-healing ability of the service failure is disabled by default. Manual operation is required to start each service one by one, and the service integration code needs to be modified to define fault self-healing related information. The operation process is complicated and the maintenance cost of the service integration code is high; therefore, a more reliable or effective solution needs to be provided

Images