Lns linac control system: Status report

Yoshinobu Shibasaki, M. Mutoh, Hiroyuki Hama · 2005

In Laboratory of Nuclear Science Tohoku University, the electron linear accelerator control system was renewed to the system that COACK [1] was implemented as kernel in 2001. At first, there were any troubles in the system. Afterward by having corrected client program and tuned COACK server, instability disappeared, and the system has been working well. This paper reports some correction points that remove malfunction of the system, information about using Windows 2003 Server as COACK system, and extension of control system by improvement of LNS Linac. TROUBLE ON COACK Trouble that caused instability to the system did not occur when the system began operation. However, communications traffic to the server increased in amount because a lot of devices were installed after that. Infrequently, it caused the following problems to the system. • Extreme increase in number of ReplyController. • Mismatch of the session ID between Clients and Server. First problem has occurred when a large amount of reply command in short time saved from one client. When a large amount of reply command in short time saved, number of executed ReplyController exceeds a limitation. In this case, COACK executes plural handling components of necessary number to process. This component is called ReplyController. However, the phenomenon that execution number reaches at a burst to the limitations was confirmed. In this condition, when COACK was stopped, components which not stopped was in the server. Second problem, with relation to the first problem, has occurred when a large amount of command from some clients was generated. This phenomenon remained a rare event, and it is thought that generating a number of commands is involved in both problems. When the second problem occurred, impossible client program to reexecute remained because the access point of session which managed by the server changed to the different ID. For recovering of the problem, the server need reboot, and it influences to all over the control system. We were able to know that to prevent it by ingenuity of operation is possible.

Read the paper · More papers on PaperTik