JPH0432936A - Tracing method for system operation in on-line service - Google Patents

Abstract

PURPOSE:To facilitate the investigation of the cause of failure and its repair by stacking trace information showing trace points are passed in a main storage when a CPU executes instructions at the trace points set previously in an OS, a package, and a transaction program during service operation. CONSTITUTION:The trace points are set before and behind a program 3 where the presence of the cause of failure is estimated and when the CPU passes the trace points in on-line processing, information on the passage is stacked in a trace point storage area on the main storage. If system failure occurs and the system is shut down, main storage dump information is gathered on a magnetic MT by the operation rule of the system. Therefore, the operation state of the program 3 right before the system failure occurs is made clear with the trace point storage area in the dump information of the main storage. Consequently, the investigation of the cause of the failure and its repair are facilitated.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a system operation tracing method during online service, and particularly to searching for the cause of a system failure in an online system.

[Conventional technology]

Conventionally, trace information of system operations has been collected on secondary media such as LP (line printer), MT (magnetic tape), and DK (magnetic disk).

Trace information of system operations is useful information for searching for the cause of system failure in large-scale online systems, but DS (
dynamic step), write time is online
This information could not be collected during online service due to the impact on system turnaround and throughput.

[Problem to be solved by the invention]

Traditionally, to collect trace information on system operations, it was necessary to conduct reproduction tests during service operations such as Saturdays and Sundays.
It took a lot of time and effort.

Furthermore, even if reproduction tests were conducted, the failures could not be reproduced due to the difference between the reproduction test environment and the actual service environment, and the effort was often in vain.

Then, while it takes time to search for the cause of the failure and repair it, there is a problem in that the same failure occurs again during the online service.

The present invention has been made in view of these points, and its purpose is to provide a system operation tracing method during online service that facilitates searching for and repairing the cause of a failure.

[Means to solve the problem]

In order to achieve such an object, the present invention provides information indicating that when the CPU executes an instruction on a trace point preset in the OS, puff cage, or business program during service operation, the trace point has been passed. The trace information is stacked on the main memory.

[Effect]

In the present invention, when the same failure occurs again, highly accurate information for searching for the cause of the failure can be obtained.

〔Example〕

First, an outline of an embodiment of the system operation tracing method according to the present invention will be described.

In this embodiment, trace points are set in advance before and after the program where the cause of the failure is expected to exist.
When the CPU passes a trace point during online processing, the information is stacked in the trace point storage area on the main memory. When a system failure occurs and the system goes down, dump information from the main memory is taken to the magnetic MT according to system operation rules. The operating state of the program immediately before the system failure can be determined from the trace point storage area in the main memory dump information.
It becomes easier to identify the location of the failure.

FIG. 1 is a block system diagram for explaining an embodiment of a system operation tracing method during online service according to the present invention. In the figure, 1 is a processing device, 2 is a program under test (O3, package, business program), 3 is a test program whose details are shown in FIG. 2, 4 is a business terminal, 5 is a communication line, and 6 is a peripheral device. It is.

In FIG. 1, trace points are set in advance in the O81 package and business program. When the CPU executes the set trace point, control is transferred to the interrupt processing program due to a hardware interrupt.The interrupt processing program stacks the trace point information in the trace point information storage area. When the trace point information storage area becomes full, it is overridden and used from the beginning.

FIG. 2(a) is a flowchart showing the trace point setting process. First, the instruction at the designated trace point is saved in the trace point table shown in FIG. 3 (step If). Then, the instruction at the designated trace point is corrected to invalid code (step 12).

Fig. 2 (Figure 2) is a flowchart showing interrupt processing. First, an address is obtained from a PSW (program status word) (step 21).

Next, write the address into the stack area (step 22
)0 Next, the instructions saved in the trace point table are executed (step 23).

When the trace point information storage area shown in FIG. 4 is full, the pointer is returned to the beginning.

〔Effect of the invention〕

As explained above, the present invention allows the OS to
,PanCage,When the CPU executes an instruction on a trace point,preset in a business program, trace information indicating that the,point has been passed is stacked in the main memory, so that,the trace information is stored during online service. Therefore, when the same failure occurs again, highly accurate information for searching for the cause of the failure can be obtained, making it easier to search for the cause of the failure and repair it.

Furthermore, by stacking the trace information on the main memory, the effect on the throughput and turnaround of online services is negligible and can be ignored.

Generally, when a system failure occurs and the system goes down, the cause of the failure occurs immediately before the system goes down. Therefore, the trace information stack area can be overridden, so the stack area can be kept small and the impact on online processing can be ignored.

[Brief explanation of the drawing]

FIG. 1 is a block diagram for explaining an embodiment of the system operation tracing method during online service according to the present invention, FIG. 2 is a flowchart showing the contents of a test program, and FIG. 3 is a trace point table. FIG. 4 is an explanatory diagram showing a trace point information storage area. Patent applicant Nippon Telegraph and Telephone Corporation NTT Software Corporation

Claims (1)

[Claims] In an online processing system where many terminals are connected via communication lines, when the CPU executes an instruction on a trace point preset in the OS, package, or business program during service operation, that point A system operation tracing method during an online service, characterized by stacking trace information indicating that the system has passed through the main memory.