A method and apparatus for monitoring a learning task

Abstract

The invention discloses a method and a device for monitoring a learning task. The method comprises the following steps: after a learning task is created, obtaining an identification of the created learning task; reading information related to the learning task from the collected and stored monitoring information of the resource using an identification of the learning task during a life cycle of the learning task; processing the read information to obtain the monitoring information of the learning task and storing the monitoring information in the database; wherein the monitoring information ofthe learning task comprises real-time information and statistical information of the learning task; the lifecycle of the learning task refers to a period of time from the start of the learning task to the end of the learning task. The monitoring method of the learning task of the embodiment of the invention can effectively monitor the task, more intuitively observe the running situation of the task, and is beneficial to the operation and maintenance of the system.

Description

Technical field [0001] The present invention relates to the field of computer technology, in particular to a method and equipment for monitoring learning tasks. Background technique [0002] In the era of artificial intelligence, machine learning is used in more and more fields, learning tasks often require a lot of calculations, and the use of distributed learning tasks greatly reduces the calculation time. A specific implementation method is to use kubernetes (container cluster management platform) to manage the cluster and create distributed learning tasks. Specifically, when creating a learning task, a kubernetes Job (a way for kubernetes to organize containers) is created. [0003] The running Job task will create containers on multiple nodes. These containers will be responsible for computing tasks and disappear after running. Currently, the method of monitoring a kubernetes cluster can only monitor the node information of a single node and the information of the containers...

AI Technical Summary

Problems solved by technology

At present, the method of monitoring the kubernetes cluster can only monitor the node information of a single node and the information of the container running on the node, but it cannot be monitored at the task level. It is more difficult to monitor tasks that will disappear, such as jobs.

Images