Research Of AST3 System High Availability

AST3 are three Schmidt Telescopes to be deployed in Kunlun Station, DOME A area of Antarctica continent for astronomical observation and study of variable stars. Kunlun Station still lacks the ability to function as an overwintering station. As a result, AST3 Software System should be able to work continuously for up to six months under unattended condition.This paper focuses to improve reliability and availability of AST3 system in the following areas: per-node reliability, cluster reliability and data backup.In node level, AST3 system is divided into photometry process and daemon process, and the latter will create worker processes when new images are detected. The daemon will keep monitor the processes and take care of them once they go abnormal. Besides, dynamic reconfiguration of runtime arguments, recovery after crash, logging system is supported. Cross protection and Cron protection is used to assure high availability of daemon.In cluster level, data processing machines of the telescopes are well organized to compose a high-availability cluster. Resources are unified dispatched to provide processing services. All the services will still be available even some of nodes go down.As for data backup, according to different importance and recovery hardness, scientific data produced by AST3 is well contained using different backup strategy and devices. The data will still be available even after disasters.