JP2002049509A - Data processing system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the reliability by duplexing the firmwares of a data processing system composed of plural nodes. SOLUTION: Firmwars 14A and 14B which differ in version are stored in ROMs 12A and 12B. A monitor device 18 monitors the processing state of the node 10 or 20 that it belongs to by a CPU 11. Respective monitor devices when detecting trouble during firmware execution communicate with each other to switch the ROMs and perform restarting. The respective nodes can synchronously switch the firmwares to use through a series of the operations. If a trouble occurs, restarting is performed by a firmware of safe version, and whichever node has starting trouble, it is guaranteed that all the nodes use firmwares of the same version.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data processing system for controlling information processing and equipment, and more particularly to a data processing system comprising a plurality of nodes capable of reducing downtime and improving reliability. It relates to a data processing system.

[0002]

2. Description of the Related Art In recent years, there has been an increasing demand for high reliability of data processing systems. In particular, in a data processing system used as a server, there is a problem of reducing downtime due to a failure. Generally, the reliability of hardware failures is ensured by using hardware duplication or degraded operation.However, failures due to program bugs should be resolved using a similar approach. Can not.

In order to cope with a failure caused by such a program such as software or firmware, the system is provided with a plurality of programs, and the operation of the system is continued by switching and executing the program to be executed when the failure occurs. And the reliability can be improved.

As a prior art capable of coping with a failure caused by the above-described program, for example, Japanese Patent Laid-Open No.
A technique described in, for example, JP-A-1-154099 is known. This conventional technique stores a plurality of firmwares represented by different codings in a single system in different ROMs, and changes settings so that when a failure is detected during execution of the firmware, different firmware is executed, This is to restart the system. According to such a conventional technique, it is possible to recover a failure by preparing a firmware that does not include a bug that causes a failure (for example, an old version of firmware) as the firmware to be restarted. it can.

[0005]

As described above, the data processing system is required to have high reliability, but also to have higher operation performance. In order to solve this, a high-performance server is often configured in a multi-node configuration in which a plurality of nodes having a certain performance are connected via a connection device. When the number of nodes configuring the system increases, it becomes difficult to share a single firmware.
A copy of the firmware is located on each of the plurality of nodes. In general, since the firmware executed in the system must be a single firmware, when performing firmware duplication in the above-described system, it is necessary to control all nodes to use the same firmware. No.

In the above-mentioned prior art, no consideration is given to the case where the data processing system is composed of a plurality of nodes. When this prior art is applied to a multi-node system, each node becomes independent. As a result, there is a problem that the firmware used in the node that has succeeded in starting and the firmware used in the node that has failed to start do not match.

SUMMARY OF THE INVENTION An object of the present invention is to solve the above-mentioned problems of the prior art, to reliably recover from a failure caused by a bug in a program executed in a data processing system including a plurality of nodes, and to improve reliability. Another object of the present invention is to provide a highly reliable data processing system.

[0008]

According to the present invention, there is provided a data processing system comprising a plurality of nodes connected to each other by a communication device, wherein the node comprises one or more processors for executing a program. Program storage means for holding at least two programs, program selection means for selecting one of the programs to be executed by the processor, and non-volatile memory for holding information indicating programs to be executed with priority And a processing failure determination means for detecting a failure during execution of the program, wherein the processing failure determination means, when detecting a failure in the program being executed, all processing failure determination means belonging to the data processing system Notify program failure in own node and execute using program selection means of own node And selecting another program that is not currently being executed in place of the program and causing the processor to execute the program. When a failure is notified from at least one other node, the currently executed program is executed instead of the currently executed program. This is achieved by selecting and running other programs that are not.

In the above solution, each of the two programs may be a new version of the program,
And a previous version of the program, and these programs may be firmware stored in a read-only memory or software stored in a non-volatile recording medium.

[0010]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of a data processing system according to the present invention will be described below in detail with reference to the drawings.

FIG. 1 is a block diagram showing an overall schematic configuration of a data processing system according to one embodiment of the present invention, and FIG. 2 is a flowchart for explaining processing at the time of starting the data processing system according to one embodiment of the present invention. . 1, 10 and 20 are nodes, 11 is a CPU, 12A and 12
B is a ROM, 14A and 14B are firmware, 15 is a switching means, 16 is a system bus, 17 is a non-volatile memory, 18 is a monitoring device, and 30 is a connection device.

A typical example of a data processing system according to an embodiment of the present invention is a server computer, and as shown in FIG.
And a connection device 30 that connects them and performs communication between nodes. Nodes 10, 20
Have the same configuration, a central processing unit (CPU: Central Processing Unit) 11 which is a processor, a monitoring device 18 which is a processing failure determining means for detecting a failure when executing firmware, and a non-volatile memory 17. Are connected to a system bus 16. The system bus 16 further has two read-only memories (RO).
M) 12A and 12B are connected via the switching means 15. The figure shows two nodes 10,
Although only 20 is shown, more nodes may be provided. Further, a plurality of CPUs may be provided in the node.

The ROMs 12A and 12B are program storage means for storing and holding firmware A and B (14A and 14B) whose contents are partially different from each other. The switching means 15 selects any one of the ROMs storing the firmware as a firmware reading destination,
Connect to system bus 16. That is, the switching unit 16 functions as a program selection unit. Note that R
OM12A and 12B are EPROM and flash ROM
It may be an erasable read-only memory such as

In the nonvolatile memory 17, information for identifying a ROM to be started preferentially is written. For example, the identification numbers “1” are stored in the ROMs 12A and 12B, respectively.
Assuming that “2” is given, when the firmware of the ROM 12A should be started with priority, “1”
"2" when priority should be given to 2B firmware
Is written. This value is retained even when the power of the data processing system is turned off, and is referred to at the next startup.

Next, the operation of the data processing system according to the embodiment of the present invention configured as described above will be described with reference to the flow shown in FIG. The example shown in FIG.
It is assumed that the firmware 14A, 14B is a boot program, and the flowchart describes the process when the data processing system is started. And firmware 14A, 14B
Are the new version of the firmware and the previous version of the firmware, respectively.

(1) When the data processing system is started, the switching means 15 acquires the ROM identification information held in the non-volatile memory 17 and selects the ROM to be started first from the ROMs 12A and 12B. And that RO
Activate the processor using the firmware in M.
Here, the ROM to be booted preferentially is the RO that stores the updated new version of firmware.
It is assumed that M is present (steps 100 to 120).

(2) The monitoring device 18 monitors the execution state of the firmware and checks whether a failure has occurred in the execution of the firmware and the activation of the processor has failed (step 130).

(3) If it is detected in step 130 that the activation of the processor has failed, the monitoring device 18
An error is reported to the monitoring devices of all other nodes, and the processor of the own node is stopped. The monitoring device 18
The method of monitoring a failure is well known and will not be described (steps 150 and 160).

(4) If it is detected in step 130 that the activation of the processor is successful, the monitoring device 18
It is checked whether or not an abnormality has been reported from the monitoring device of another node, and if there is no abnormality notification, the startup processing is completed as it is (step 140).

(5) If an abnormality is reported from the monitoring device of another node in step 140, the processor of the own node is stopped (step 160).

(6) In step 160, after the monitoring device 18 stops the processor, the switching means enables the ROM on the side not selected at the time of startup and uses the old version firmware before updating to the processor. To restart. By restarting the processor, the CPU 11
Starts execution of firmware different from the last boot. At this time, since firmware switching has occurred in all nodes of the system, all processors are executing the same firmware (step 17).
0, 180).

(7) The monitoring device 18 monitors the execution of the firmware again, and checks whether a failure has occurred in the execution of the firmware and the activation of the processor has failed (step 190).

(8) If it is detected in step 190 that the activation of the processor has failed, the monitoring device 18
An error is reported to the monitoring devices of all other nodes, and the processor of the own node is stopped (steps 210 and 2).
20).

(9) If it is detected in step 190 that the activation of the processor is successful, the monitoring device 18
It is checked whether an abnormality has been reported from the monitoring device of another node, and if there is no abnormality notification, the startup processing is completed as it is (step 200). (10) Step 220
In the case where the failure occurs again and the processor is stopped even when the firmware is switched and executed, the possibility of a hardware failure is high. Therefore, the monitoring device 18 immediately stops the system and takes a measure by the user. wait. The monitoring device 18 can be configured to notify the user of such a fault (step 230).

In the above-described embodiment of the present invention, the processing operation is described assuming that the firmware is a boot program, the firmware is stored in the ROM, and the data processing system is started.
The present invention is not limited to the above-described firmware, but can be applied to other software such as an application program recorded on a fixed magnetic disk device, an optical disk device, a magnetic tape, or the like. In this case, a set of software consisting of two versions, old and new, is stored in a disk device or the like, and the software is switched and started in the event of a failure. Restarting with the old version can be performed, and as a result, downtime can be reduced.

According to the above-described embodiment of the present invention, in a data processing system including a plurality of nodes each having a program storage means in each node, the monitoring device installed in each node performs communication, By controlling the switching means of the node to select the same program, the program is duplicated,
The reliability of the system can be improved.

According to the data processing system according to the embodiment of the present invention described above, one of the programs to be duplicated is a new version, and the other is an old version. When the state falls, the system can be automatically and promptly restarted by using the old version of the program, so that the down time of the system can be reduced.

[0028]

As described above, according to the present invention, it is possible to reliably recover from a failure caused by a bug in a program executed in a data processing system including a plurality of nodes, and to obtain highly reliable data. A processing system can be provided.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an overall schematic configuration of a data processing system according to an embodiment of the present invention.

FIG. 2 is a flowchart illustrating a process at the time of starting the data processing system according to the embodiment of the present invention.

[Explanation of symbols]

 10, 20 node 11 CPU 12A, 12B ROM 14A, 14B Firmware 15 Switching means 16 System bus 17 Non-volatile memory 18 Monitoring device 30 Connection device

 ──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 5B018 GA04 HA04 HA05 HA32 MA23 NA04 QA20 5B042 GA12 JJ04 KK02 5B045 JJ42 JJ48 5B076 AB19 CA01 CA05

Claims (3)

[Claims] 1. In a data processing system comprising a plurality of nodes connected to each other by a communication device, said node includes one or more processors for executing a program and program storage means for holding at least two programs. Program selecting means for selecting one of the programs and causing the processor to execute the program, a non-volatile memory holding information indicating a program to be executed with priority, and a process of detecting a failure at the time of executing the program A failure determination unit, wherein when the processing failure determination unit detects a failure in the running program, the processing failure determination unit notifies all the processing failure determination units belonging to the data processing system of the program failure in the own node. Not currently executing in place of the program being executed using the program selecting means Selecting another program to be executed by the processor and, when notified of a failure from at least one other node, selecting and executing another program that is not currently executed in place of the currently executed program; A data processing system characterized by the following. 2. The data processing system according to claim 1, wherein the program is firmware stored in a read-only memory or software stored in a nonvolatile recording medium. 3. The data processing system according to claim 1, wherein each of the two programs is a new version of the program and a previous version of the program.