JPS6383843A - System for collecting trace information - Google Patents

Abstract

PURPOSE:To quickly and accurately cope with the occurrence of a fault by continuing the replacement of trace buffers only even while the collection of trace information is halted. CONSTITUTION:A trace control part 5 receives a command for the stop of a tracing action from the outside of a system and gives a command to only a trace file writing part 3 to stop its working. That is, the working of a trace buffer writing part 1 is continued as it is even though the part 3 stops its working. As a result, the receding is never carried out to a trace file although a trace buffer 2 is filled and the latest trace information is successively written on the oldest trace information. Thus only the latest trace information equal to the capacity of the trace buffer 2 is left at the buffer 2. A fault information file writing part 10 writes the latest fault information written into a fault information buffer 7 and its related peripheral trace information to a fault information file 11.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to a method for collecting trace information used in electronic computer systems.

In conventional technical computer systems, appropriate operational information is collected as trace information in order to be used for investigating the cause of failures that occur within the system.

Conventionally, to collect trace information as described above, it is first written to a small-capacity trace buffer, and whenever this trace buffer is almost full, its contents are written to a trace file on a large-capacity secondary storage device. is written in.

Problems to be Solved by the Invention In the conventional trace information collection method described above, writing of trace information to a trace buffer and writing to a trace file when the buffer is full are combined as an inseparable operation. That is, if collection of trace information is unavoidably stopped in order to reduce the load on the computer system, writing of trace information to the trace buffer will also be stopped. Therefore, if a failure such as a system down occurs in such a state, there is a problem in that trace information immediately before the failure occurs, which is necessary for investigating the cause, becomes completely impossible to obtain.

In addition, in parallel with the collection of trace information to the trace file described above, failure information is collected by nearby tracing, which collects the contents of the trace buffer only in the vicinity of the time of occurrence of the failure each time a specified failure occurs in the computer system. A collection method may be adopted. However, in the conventional trace information collection method, updating of the trace buffer is also stopped when the trace operation is stopped, so there is a problem in that the above-mentioned collection of failure information by nearby tracing becomes useless.

Structure of the Invention ``Means for Solving the Problems'' The trace information collection method of the present invention is such that while the collection of trace information to a trace file is stopped in order to reduce the load or improve the performance of a computer system, By continuing to update the trace buffer, even in the event of a system failure, you can secure at least the trace buffer's capacity at least as much previous trace information as necessary to investigate the cause of the problem, and also prevent problems caused by nearby traces. Configured to enable information collection at all times.

In other words, the trace information collection method according to the present invention is as follows:
A writing section to the trace buffer that constantly writes trace information generated within the computer system to the trace buffer by overwriting old written data when there is no free space for writing, and trace information received from outside the system. A trace file writing unit that starts and stops the operation according to the collection start/stop command, and writes the contents to the trace file each time the trace buffer becomes full during operation, and a trace file writing unit that writes the contents to the trace file each time the trace buffer becomes full. A writing section to the fault information buffer writes information to the fault information buffer, and when a predetermined fault occurs, the trace information in the vicinity is read from the trace buffer and written to the fault information file together with the fault information read from the fault information buffer. It is equipped with a disability sentiment collection department.

Hereinafter, the operation of the present invention will be explained in detail together with examples.

Example FIG. 1 is a conceptual diagram of an embodiment of the present invention.

In the figure, 1 is a section that writes to the trace buffer, 2 is a trace buffer, 3 is a section that writes to a trace file, 4 is a trace file, 5 is a trace control section, 6 is a section that writes to a fault information buffer, and 7 is a fault An information buffer, 8 a failure type determination unit, 9 a nearby trace unit, 10 a failure information file writing unit, and 11 a failure information file.

The trace buffer writing unit 1 writes corresponding trace information to the trace buffer 2 every time an event that requires the collection of trace information occurs in a computer system to which this method is applied.

When the trace file writing unit 3 operates,
Each time the trace buffer 2 becomes full, its contents are written to the trace file 4.

When the trace control unit 5 receives a trace stop command from outside the system, the trace control unit 5 writes the trace file writing unit 3
It only commands the start and stop of the above operations. That is,
Even if the trace file writing unit 3 stops operating, the trace buffer writing unit 1 continues to operate. As a result, even if trace buffer 2 becomes full, it is not saved to the trace file, and the latest trace information is sequentially overwritten by the oldest trace information, so that the latest trace information equal to the capacity of the trace buffer is Only the trace data will be left in the trace buffer 2.

In parallel with this, a failure information collection unit consisting of parts numbered 6 to 11 collects failure information with nearby trace information for a predetermined failure that has occurred within the computer system. Collection takes place.

That is, each time a fault occurs in the computer system, the fault information buffer writing unit 6 writes information regarding the fault, including the fault code and time of occurrence, to the fault information buffer 7, and then terminates the writing. is notified to the failure type determination unit 8.

The failure type determination unit 8, which received the notification of the end of writing,
By reading the latest fault information written in the fault information buffer and comparing the fault code contained therein with a predetermined fault code, it is determined whether the fault requires nearby tracing, and the necessary If so, nearby trace section 9
Start.

The activated nearby trace section 9 successively outputs a predetermined number of items of nearby trace information from the trace buffer 2 and writes it into the fault information buffer 7, and then notifies the writing section 10 of the completion of this writing to the fault information file. Notify. The failure information file writing unit 10, which has received the notification of the completion of writing, writes the latest failure information written in the failure information buffer unit and related nearby trace information to the failure information file 11.

Effects of the Invention As explained in detail above, the trace information collection method according to the present invention allows only updating of the trace buffer even when the collection of trace information is stopped in order to reduce the load on the computer system or improve its performance. Since this is a continuous configuration, it is possible to secure at least the capacity of the trace buffer for the immediately preceding trace information needed to investigate the cause of a failure such as a system down, and the effect is that failures can be dealt with quickly and accurately. .

Furthermore, according to the present invention, since detailed failure information is collected for a given failure by also collecting failure information including nearby trace information, it is possible to deal with the occurrence of a failure more quickly and accurately. The effect is produced.

[Brief explanation of the drawing]

FIG. 1 is a conceptual diagram of a trace information collection method according to an embodiment of the present invention. 1... Writing part to trace buffer, 2... Trace buffer, 3... Writing part to trace file, 4... Trace file, 5... Trace control part, 6... Failure information 7. Fault information buffer, 8. Fault type determination section, 9. Nearby trace section, 10. Writing section to fault information file, 11. Fault information file.

Claims (1)

[Scope of Claims] A trace buffer writing unit that constantly writes trace information in a computer system to the trace buffer by overwriting old written data when there is no free space for writing; a trace file writing unit that starts and stops operation according to the received trace information collection start and stop commands, and writes the contents to a trace file each time the trace buffer becomes almost full during operation; a fault information buffer writing unit that writes information regarding a fault that has occurred in the fault information buffer; A method for collecting trace information, comprising: a fault information collection unit that writes information together with a fault information file.