US 11,809,303 B2
High-performance computing-oriented method for automatically deploying execution environment along with job
Kai Lu, Hunan (CN); Wenzhe Zhang, Hunan (CN); Ruibo Wang, Hunan (CN); Wanqing Chi, Hunan (CN); Yong Dong, Hunan (CN); Wei Zhang, Hunan (CN); Huijun Wu, Hunan (CN); and Mingtian Shao, Hunan (CN)
Assigned to NATIONAL UNIVERSITY OF DEFENSE TECHNOLOGY, Hunan (CN)
Filed by NATIONAL UNIVERSITY OF DEFENSE TECHNOLOGY, Hunan (CN)
Filed on Nov. 26, 2021, as Appl. No. 17/535,702.
Claims priority of application No. 202011395015.4 (CN), filed on Dec. 3, 2020.
Prior Publication US 2022/0179774 A1, Jun. 9, 2022
Int. Cl. G06F 11/36 (2006.01); G06F 9/445 (2018.01); H04L 9/40 (2022.01)
CPC G06F 11/3656 (2013.01) [G06F 9/44505 (2013.01); G06F 11/366 (2013.01); H04L 63/083 (2013.01)] 8 Claims
OG exemplary drawing
 
1. A high-performance computing-oriented method for automatically deploying an execution environment along with a job, comprising:
S1) presetting a plurality of isolated execution environments at nodes of a high-performance computing system based on an overlay file system and an isolation process, wherein the plurality of isolated execution environments are configured to implement sharing an environment for users and to implement a customization environment defined by each of the users and the automatic deployment of the execution environment defined by the each of the users;
S2) launching an isolated lightweight development environment corresponding to the user in response to the user logging in at a login node of the high-performance computing system;
S3) receiving a development and debugging operation on the job of the user and a configuration of the execution environment on the job of the user executed by the user at the login node, and issuing a job executing request to a job management system;
S4) assigning one of compute nodes from the nodes of the high-performance computing system for the job of the user, loading the job, and automatically deploying an upper file system corresponding to the user synchronously to the assigned compute node along with the job through the job management system, wherein the job of the user is executed at the assigned compute node; and
S5) identifying one or more executing results generated when the job of the user is executed, and transmitting the one or more results to the login node at which the user is logged in by the assigned compute node after the job is executed, clearing the upper file system corresponding to the user at the assigned compute node, and ending the job and the isolation process.