Operation retrying method and device of task

Abstract

Embodiments of the invention provide an operation retrying method and device of a task. The method comprises the following steps of: detecting whether an operation in the task fails or not; if the detection result is positive, carrying out statistic on the progress of the task; calculating a retrying time of the progress of the task; and after waiting for the retrying time, re-executing the operation. According to the method and device provided by the embodiments of the invention, the retrying time is adaptively calculated according to the progress of the task when the operation of the task fails, so that the length of the retrying time can be greatly improved for long tasks, dynamic task retrying can be realized, the service interrupt condition with longer time can be handled, the equipment resource waste caused by re-executing the tasks during task failure is avoided, and the retrying cost is greatly reduced when the success rate of the tasks is ensured.

Description

technical field [0001] The present application relates to the technical field of computer processing, in particular to a job operation retry method and a job operation retry device. Background technique [0002] In a computer system, a job will interact with other systems (services) during operation. When other systems (services) are unavailable for a short period of time (service interruption), the job will fail to run. [0003] In the big data scenario, when the current product runs wrong, it generally retries the job, that is, waits for other systems (services) to recover, so that the job can continue to run, and the success rate of the job can be guaranteed as much as possible. [0004] Otherwise, if all retries fail, the entire job will fail. [0005] For example, if other systems (services) are unavailable for a maximum of 10 minutes, the overall retry interval is required to be greater than 10 minutes to ensure that the job can continue to run when such a short-term ...

AI Technical Summary

Problems solved by technology

[0007] 1. Short jobs; for example, for a job with an estimated running time of only 30 minutes, the actual running time due to retries may be longer than the normal running time;

[0008] 2. When the administrator needs to manually stop the job, especially when the external system (service) is unavailable, it is necessary to manually stop the job for operation and maintenance, and make some adjustments and deployments. If the job retry happens to happen, it needs to wait for the job to retry After 10 minutes, the job can be completely stopped

[0011] However, this solution may lead to increased retry costs for long jobs

[0012] For example, for a job that runs for 10 hours, if the 10-hour job runs to 80%, that is, when it has been running for 8 hours, a 10-minute service interruption occurs, exceeding the maximum retry time, resulting in The job fails, and re-executing the job means that the previous 8 hours of running are wasted, at a huge cost

Images