JPH0827768B2 - Data processing device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Outline] A data processing device using a kernel of an operating system is provided with a transmission space and a plurality of reception spaces for the purpose of lightening a burden on a person who performs message communication for realizing a fault tolerant function. In the mediating means of the data processing device for transmitting and receiving messages between the two via the mediating means, when an abnormality is detected in one of the storage means and the receiving means for storing the request message from the transmission space to the reception space, Retransmitting means for retransmitting the request message addressed to the receiving space in which the abnormality has occurred and the identification information received from the transmitting space before detection to another receiving space and storing the identifying information, and from the other receiving space. Receiving means for receiving the identification information together with the reply message to the request message, and the reply message of the reply message. Determining means for determining whether or not the different information and the identification information of the request message match, and if the determination result is the same, the reply message is transmitted to the transmission space, and if the determination result is not the same, It is configured to include a reply control unit that causes the retransmission unit to perform the retransmission and stops the process when the retransmission satisfies a predetermined criterion.

[Industrial applications]

The present invention relates to a data processing device, and more particularly to a data processing device using an operating system kernel.

Data is transmitted and received in the data processing system in the form of messages. Therefore, with regard to the transmission and reception of this message, a fault tolerant (FT) function, that is, a function of continuously providing the service to the network without stopping even if an abnormality occurs in the system is realized.

[Conventional technology]

FIG. 4 is an explanatory view of a conventional technique and shows a conventional message retransmission process.

In FIG. 4, 1 is a transmission space, 2 is a kernel, 3 is a reception space, 31 is an active space in the reception space 3, and 32 is a standby space in the reception space 3.

A message (request message) that the sender, such as a processing program or process, requests the processing by the sending space 1.
Via the kernel (core) 2 of the operating system, the receiving side such as another processing program or process, that is, the currently active space of the receiving space 3 (active space 3
Send to 1) (in the figure).

After that, if an abnormality such as a crash occurs in the active space 31 during message processing (in the figure), the kernel 2 monitoring the active space 31 notifies the transmit space 1 of the occurrence of the crash, and the transmit space 1 Receives this (in the figure).
Further, due to the occurrence of the crash, the active space 31 is switched to the standby space 32, which is the reception space 3 paired with this (duplex), and is active (in the figure).

The transmission space 1 which has learned the occurrence of the crash retransmits the request message via the kernel 2 to the standby space 32 which is the currently active reception space (in the figure).

The standby space 32 receives the request message and completes the processing, and at the same time, sends a reply message to the transmission space 1 via the kernel 2 (in the figure), and the transmission space 1 receives this (in the figure). .

As described above, the transmission space 1 mainly realizes the fault tolerant function against the crash in the working space 31.

If the cause of the crash in the active space 31 is a bug in the processing program on the receiving side such as an input-dependent bug, the same crash occurs in the standby space 32 and normal processing cannot be performed. As a result, a crash or an infinite loop in which request messages are retransmitted may occur. Therefore, loop detection processing is performed before retransmission (in the figure), and when the number of loops exceeds a predetermined value, appropriate processing is performed to eliminate the loop.

Therefore, the fault tolerant function for the loop is realized by the transmission space 1.

[Problems to be Solved by the Invention]

According to the above-mentioned conventional technique, the transmission space 1 must perform the process (in the figure) of recognizing the switching of the receiving space based on the crash and the notification of the crash and retransmitting the request message. On the other hand, the content of the retransmitted request message must be managed by the side using the request message, that is, the receiving side (reception space 3). Therefore, the receiving side must have a function of checking whether the content of the request message received after the second time is the same as the first one. Further, as described above, a loop may occur, and therefore the transmission space 1 needs to perform loop detection for this.

As described above, conventionally, the fault tolerant function for message transmission / reception has been realized by a person who performs message communication, that is, the transmission space 1 and the reception space 3 (processing program or process). That is, from the viewpoint of a person who performs message communication, especially from the transmission space 1, the request message for fault-tolerant is not automatically retransmitted. For this reason, there is a problem that a person who performs message communication has a heavy load for realizing the fault tolerant function.

It is an object of the present invention to provide a data processing device that reduces the burden on a person who performs message communication for realizing a fault tolerant function.

[Means for solving the problem]

FIG. 1 is a principle block diagram of the present invention, showing a data processing device according to the present invention.

In FIG. 1, 1 is a transmission space, 2 is a kernel, 3 is a reception space, 21 is an automatic re-transmission processing unit, 22 is a reply message monitoring processing unit, 23 is a loop detection processing unit, 31 is an active part of the reception space 3. Space 32 is a standby space in the reception space 3.

The transmission space 1 transmits a request message (a message requesting processing) to the reception space 3 via the kernel 2.

When receiving the request message, the reception space 3 performs the processing requested by the request message. Then, when the processing is completed, the reception space 3 transmits a reply message to the request message to the transmission space 1 via the kernel 2. This notifies the completion of processing. The reception space 3 is duplicated in preparation for the occurrence of an abnormality such as a crash, and is composed of an active space 31 which is a space currently active and a standby space 32 which is a standby space. If an abnormality occurs in the working space 31 of the receiving space 3, the receiving space 3 becomes the working space.
After switching from 31 to the standby space 32, the request message is retransmitted to the standby space 32, which is the receiving space 3, and the requested processing can be completed.

The kernel 2 performs a service regarding message communication such as replying to the request message and the reply message described above, and also performs an automatic resending process of the request message. For this reason, the automatic resend processing unit 21 and the reply message monitoring processing unit 2
2 and the loop detection processing unit 23 are provided in the kernel 2.

The automatic retransmission processing unit 21 recognizes that an abnormality such as a crash has occurred in the working space 31 of the reception space 3 and that the space has been switched as described above. And
The request message transmitted from the transmission space 1 to the reception space 3 prior to the occurrence of the abnormality is retransmitted to the standby space (currently used space) 32 of the reception space 3.

The reply message monitoring processor 22 monitors reply messages. When the request message is retransmitted, the reception space 3 (active space 31 and / or standby space)
32) The correct reply message sent from
That is, only the reply message to the request message retransmitted at the closest time is selected and transmitted to the transmission space 1.

When the request message is retransmitted, the loop detection processing unit 23 detects a loop in which an abnormality such as a crash occurs in the reception space 3 or the request message is retransmitted repeatedly.

[Action]

In the kernel 2, the transmission space 1 is the active space 31 of the reception space 3
Receives the request message sent to the destination and sends it to the working space 31. At this time, the automatic resend processing unit 21 copies and saves the request message.

When an abnormality such as a crash does not occur in the working space 31, the working space 31 sends a reply message to the request message with the sending space 1 as a destination when the requested processing is completed. The reply message monitoring processing unit 22 of the kernel 2 which received this recognizes that the reply message is correct and sends it to the transmission space 1.

When an abnormality such as a crash occurs in the working space 31, the working space 31 is switched to the standby space 32.

When the automatic resending processing unit 21 recognizes that an abnormality has occurred and the space has been switched, it sends the stored request message to the standby space 32. Therefore, the request message can be retransmitted while maintaining the identity of the request message and without imposing any burden on the transmission space 1.

At the time of this retransmission, the loop detection processing unit 23 detects a loop and performs necessary processing such as loop cancellation.
As described above, since the loop detection processing unit 23 of the kernel 2 performs the loop detection as the automatic retransmission processing unit 21 of the kernel 2 performs the retransmission, the loop detection is surely performed and the load on the transmission space 1 is reduced. It can be lost.

If no abnormality such as a crash occurs in the receiving space 3 after the retransmission is performed, the retransmission is performed from the receiving space 3 (active space 31 and / or standby space 32) upon completion of the requested processing. A reply message to the request message is sent. On the other hand, depending on the timing of occurrence of an abnormality such as a crash, a reply message to the request message before the crash may be transmitted.
In such a case, the reply message monitoring processing unit 22
Selects a reply message to the request message retransmitted at the closest time and sends it to the transmission space 1. Therefore, since the transmission space 1 receives only one reply message for one request message, it is not necessary to pay attention to the abnormality of the reception space 3.

In this way, the kernel 2 retransmits the request message for fault tolerance and guarantees the identity of the request message at that time, so that the load on the transmission space 1 and the reception space 3 can be reduced.

〔Example〕

FIG. 2 is an explanatory diagram of the embodiment and shows a data processing device.

In FIG. 2, 11 is a request message transmission processing unit with automatic retransmission (request message transmission processing unit), 12 is a reply message reception processing unit, 24 is a copy area, and 33 is a reply message transmission processing unit.

The transmission space 1 and the reception space 3 are persons who perform message communication in the illustrated data processing device which is a message-based system, and are, for example, processing programs, processes, central processing units, terminals and the like.

The transmission space 1 which is the transmission side of the message communication includes a request message transmission processing unit 11 and a reply message reception processing unit 12
Is provided.

The request message transmission processing unit 11 determines whether the request message is automatically retransmitted (requested) or not automatically retransmitted according to the user's designation, and then performs the transmission process. The former is for requesting the kernel 2 to perform an automatic retransmission process for a request message. Therefore, even if a crash occurs in the working space 31 of the receiving space 3 (if the number of loops is small), the reply message can be received. On the other hand, the latter sends only the request message. Therefore, when the crash occurs, the processing request for the reception space 3 is stopped.

The reply message reception processing unit 12 receives only one correct reply message selected by the reply message monitoring processing unit 22 even when the automatic resend processing of the request message is performed. If a request message without automatic retransmission is sent, the reply message reception processing unit 12
Receives a crash notification when a crash occurs. As a result, the state having the communication message is released.

The receiving space 3 which is the receiving side of the message communication is duplicated so that even if a crash occurs in the active space 31, the standby space 32 can immediately continue the activity. Therefore, the standby space 32 has the same logical configuration as the active space 31, and the reply message transmission processing unit 33 exists in the active space 31 as well as the standby space 32. Switching between the active space 31 and the standby space 32 is performed by the kernel 2 or the operating system.

The reply message transmission processing unit 33 transmits a reply message according to the processing in the space.

The kernel 2 is the core or supervisor of the operating system.

The automatic retransmission processing unit 21 receives the request message transmitted from the request message transmission processing unit 11 and sends it to the reception space 3. At this time, it is identified whether or not this request message has automatic retransmission, and when it has automatic retransmission, the request message is copied and saved in the copy area 24 in the kernel 2. Then, after adding the number of retransmissions indicating the number of times of transmission to the request message, this is added to the reception space 3
Send to At the time of the first transmission, the number of retransmissions = 0. If the request message does not have automatic retransmission, the copying and addition of the number of retransmissions are not performed.

In addition, the automatic retransmission processing unit 21 is configured to receive the reception space 3 (active space 3
When a crash occurs in 1), the request message is extracted from the copy area 24 and is received in the receiving space 3 (standby space 3
2) Resend to. At this time, the number of retransmissions is incremented by +1 and added to the request message.

At the time of this retransmission, prior to this, the loop detection processing unit
23 performs loop detection. That is, if the number of retransmissions is 2 or more, a loop may be formed. Therefore, the loop detection processing unit 23 regards the number of retransmissions, that is, the number of loops,
It is checked whether or not this exceeds a predetermined value, and if it exceeds, a process for eliminating the loop is performed.

The reply message monitoring processing unit 22 receives the reply message transmitted from the reply message transmitting processing unit 33 and sends it to the reply message receiving processing unit 12. At this time, it is checked whether or not this reply message is correct as described above. For this purpose, the number of retransmissions in the request message is compared with the number of retransmissions in the reply message. The reply message transmission processing unit 33 adds the number of resends to the reply message (the number of resends added to the request message corresponding to the reply message), and then sends the reply message. In the case where the two retransmitted times thus obtained match (the addition of the retransmitted times corresponds to the addition of the identification information referred to in the present invention), the reply message monitoring processing unit 22 concerned The reply message is judged to be correct (for the request message retransmitted at the closest time) and is sent to the transmission space 1. If the number of retransmissions does not match, the reply message is not transmitted to the transmission space 1.

 FIG. 3 is a flow chart of the automatic message resend processing unit.

The automatic retransmission processing unit 21 receives the request message from the transmission space 1 and checks whether automatic retransmission is included.

The automatic resend processing unit 21 copies the request message to the copy area 24 when the automatic resending is included. If automatic resending is not included, copying will not be performed.

The automatic retransmission processing unit 21 transmits to the active space 31 a request message to which the number of times of retransmission is added at the time of request with retransmission.

The reply message monitoring processing unit 22 receives the crash notification from the receiving space 3 (standby space 32), and the automatic resending processing unit
21 and the loop detection processing unit 23 are notified. That is, in the case of the automatic resending, the crash notification is not sent to the transmission space 1. When the automatic retransmission is not included, the processing described later is performed.

The loop detection processing unit 23 performs loop detection processing to check whether the number of retransmissions exceeds a predetermined value.

When it does not exceed the predetermined value, the automatic retransmission processing unit 21 that has received the notification to that effect counts up the number of retransmissions by +1. Then, after this, the process, that is, the request message is retransmitted.

If it exceeds the predetermined value, the processing described later is performed. Further, the transmission space 1 receives an error message indicating that the number of loops has been exceeded.

The reply message monitoring processor 22 receives the reply message from the receiving space 3.

The reply message monitoring processing unit 22 checks whether the resend times of the reply message match.

If they do not match, the processing is terminated without performing the following processing and the reception of the reply message is performed.

If they match, it is not necessary to retransmit, and the copy area 24 is released by the reply message monitoring processing unit 22.

Even when the number of retransmissions exceeds a predetermined value in the processing, no further retransmission is performed, so the copy area 24 is released.

The reply message monitoring processor 22 receives the reply message and transmits it to the transmission space 1 which is the destination.

If the request message does not have automatic resending, the crash notification message is received when the crash notification is received, and the reply message is received when the reply message is received.

〔The invention's effect〕

As described above, according to the present invention, when an abnormality occurs in the receiving space, the request message is automatically retransmitted by the kernel without being aware of this in the transmitting space. Therefore, the fault tolerant function of message communication can be facilitated. In addition, the burden for this can be prevented from being applied to the sending space and the receiving space for message communication.

[Brief description of drawings]

FIG. 1 is a block diagram of the principle of the present invention, FIG. 2 is an explanatory diagram of an embodiment, FIG. 3 is a message automatic retransmission processing flow, and FIG. 1 is a transmission space, 2 is a kernel, 3 is a reception space, 21 is an automatic retransmission processing unit, 22 is a reply message monitoring processing unit, 23 is a loop detection processing unit, 31 is an active space in the reception space 3, 32 is reception It is a standby space in the space 3.

Claims (1)

[Claims] 1. A storage means for storing a request message from a transmission space to a reception space, in the mediating means of a data processing device for transmitting and receiving a message between a transmission space and a plurality of reception spaces via the mediating means. When an abnormality is detected in one of the receiving spaces, the request message and the identification information, which are received from the transmitting space before the detection and addressed to the receiving space in which the abnormality has occurred, are retransmitted to another receiving space and the identifying information is stored. Retransmitting means, receiving means for receiving identification information together with a reply message to the request message from the other receiving space, and determining whether or not the identification information of the reply message and the identification information of the request message match. If the judgment means and the judgment result match, the reply message is transmitted to the transmission space, and if the judgment result does not match. Wherein to perform the retransmission to the retransmission means, data processing apparatus, characterized in that the retransmission and a return control means for stopping the process when satisfying a predetermined criterion.