JP2907924B2 - Computer system - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Overview] Regarding a computer system for recovering resources, even if a resource use space or a PM in which the resource use space crashes, the monitoring / message transmission function in the core of the resource management space substitutes a recovery notification message. In order to collect resources by notifying the resource management space and recovering them, and to collect resources using a unified mechanism in the event of a resource use space / PM crash, and to eliminate the need for crash monitoring in the resource management space, multiple A PM (processor module), a resource use space provided in the PM to perform processing using resources, a resource management space provided in the PM to manage resources,
A resource-use space identifier correspondence table for managing resource use space identifiers / PM identifiers / resources, and a monitoring / message transmission function provided at the core of the PM, and a resource request from a resource use space in a certain PM The resource management space that has received the message registers the resource use space identifier and PM identifier extracted from the resource request message in the resource-use space identifier correspondence table, creates a resource, and sets its destination in the resource use space in a reply message. If the resource use space itself or an existing PM crashes for some reason while the resource use space is processing the resource of this destination for some reason, the monitoring / message transmission function detects this and notifies the resource use space. -A destination of a non-existent resource usage space is extracted by referring to the usage space identifier correspondence table, and a collection notification message to which the destination is assigned is extracted. Acting to inform the resource management space, configured to recover resources.

[Industrial applications]

The present invention relates to a computer system for recovering resources.

[Conventional technology]

Conventionally, as shown in FIG. 3, in a system in which a plurality of PMs (processor modules) are interconnected via a bus, a plurality of spaces are created in each PM, and each space performs processing independently, A space requesting to use (hereinafter referred to as a resource use space) sends a resource request message to a space that manages resources (hereinafter referred to as a resource management space), and the resource management space creates the resource and sends it to its destination (CAP, capability) and the resource use space performs various processes using this destination. Then, after the processing is completed, a collection notification message is notified to the resource management space, and a request is made to collect and collect the resources. At this time, if the resource use space crashes for some reason, the nucleus that manages the resource use space detects this, substitutes it, notifies the collection notification message to the resource management space, and collects it. Also, if the resource management space crashes for some reason, the resource management space on standby performs resource management. However, even if these fault-tolerant methods are used, if the PM that has the resource use space crashes, the collection notification message will not be sent to the resource management space, and the resources will not be collected. There was a problem.

According to the present invention, even if the resource use space or the PM in which the resource use space crashes, the monitoring / message transmission function in the core of the resource management space notifies the resource management space on behalf of the collection notification message and collects the resource use space. The purpose is to collect resources by a unified mechanism at the time of space / PM crash and to eliminate the need for crash monitoring by the resource management space.

[Means to solve the problem]

 Means for solving the problem will be described with reference to FIG.

In FIG. 1, PM1 is a plurality of PMs connected via a bus.

The resource use space 2 is a space where processing is performed using resources provided in the PM 1.

The destination 3 is a destination for using (accessing) resources.

The resource management space 4 is a space for managing resources provided in the PM 1.

The resource-used space identifier correspondence table 5 is a table for managing destination / resource used space identifier / PM identifier / resource and the like.

The monitoring / message transmission function 6 monitors the crash of the resource use space 2 and the PM 1 in which the resource use space 2 exists, and in response to the crash, sends a collection notification message to the resource management space 4.
Is to be notified.

[Action]

In the present invention, as shown in FIG. 1, the resource management space 4 which has received the resource request message from the resource use space in the PM 1
Registers the resource use space identifier and the PM identifier extracted from the resource request message in the resource-use space identifier correspondence table 5, creates a resource, notifies the resource use space 2 of the destination by a reply message, and notifies the resource use space 2 If the resource use space 2 itself or the existing PM1 crashes for some reason during processing using the resources of this destination,
The monitoring / message transmitting function 6 detects this, refers to the resource-used space identifier correspondence table 5 and notifies the resource management space 4 on behalf of the collection notification message to which the destination of the non-existent resource used space 2 is assigned, We try to collect resources.

Therefore, resource use space 2 or PM1 where this exists
Even if a crash occurs, the monitoring / message transmission function 6 in the core of the resource management space 4 notifies the resource management space 4 on behalf of the collection notification message and collects it, thereby unifying the resource use space / PM when a crash occurs. The resources can be collected by the mechanism, and the crash monitoring by the resource management space 4 can be made unnecessary.

〔Example〕

Next, the configuration and operation of one embodiment of the present invention will be sequentially described in detail with reference to FIG. 1 and FIG.

In FIG. 1, PM1 is a plurality of PMs connected via a bus, and here comprises PM-I and PM-II.

The resource use space 2 is a space where processing is performed using resources provided in the PM 1. Here, a space A and a space B exist in PM-I, and a space C exists in PM-II. The space A, the space B, and the space C that are the resource use spaces 2 are transmitted to the destination (CA
P) are notified from the resource management space 4, respectively, and access is made to these destination resources to perform processing.

The destination 3 is for using (accessing) resources. The processing is performed by accessing a control table, which is a resource, depending on the destination.

The resource management space 4 is a space for managing resources provided in the PM 1. The resource management space 4 receives a resource request message from the resource use space 2 and receives a space identifier (for example, space A) and a PM identifier (for example, PM-
I) is registered in the resource-used space identifier correspondence table 5 as shown in the figure, the destination of the created resource is registered, and the destination is notified to the resource used space by a reply message.

The resource-used space identifier correspondence table 5 is a table for managing destination / resource used space identifier / PM identifier / resource and the like.

The monitoring / message transmission function 6 monitors the crash of the resource use space 2 and the PM 1 in which the resource use space 2 exists, and in response to the crash, sends a collection notification message to the resource management space 4.
Is to be notified.

 Next, the operation of the configuration in FIG. 1 will be described.

(1) When the space A that is the resource use space 2 obtains the destination (CAP): (1-1) The resource request to which the space A is assigned the space identifier: the space A and the PM identifier where the space exists: the PM-I The message is notified to the resource management space 4.

(1-2) The resource management space 4 receiving the resource request message registers the space identifier: space A and the PM identifier: PM-I as a pair in the resource-used space identifier correspondence table 5 as shown in FIG. Register the created resource as a point and the destination for accessing this resource. Then, the reply message to which the destination is assigned is notified to the space A of the resource request source.

(1-3) The space A that has received the reply message stores the received destination (CAP), and accesses resources (for example, a control table) by this destination to perform processing.

(2) Next, referring to FIG. 2, the space A which is the resource use space 2 crashes or the PM in which the resource use space 2 exists.
A case where -I crashes will be described.

Detect a crash. This is a monitoring / message transmission function 6 in the core of the PM-II where the resource management space 4 exists.
Detects that the space A which is the resource use space 2 has crashed or the PM-I in which the space A exists has crashed.

Detect nonexistent space. This is because the monitoring / message transmitting mechanism 6 searches the resource-used space identifier correspondence table 5 for the space or PM for which a crash has been detected in the space or PM, for example, the space when the crash of the space A is detected. A, when the crash of PM-I is detected, the spaces A and B existing in the PM-I are taken out.

Send a collection notification message. This sends a collection notification message to which the space retrieved (detected) by the non-existing space is assigned a destination used.

The resource management space 4 that has received the collection notification message releases the resource of the notified destination and deletes the corresponding entry from the resource-used space identifier correspondence table 5.

As described above, when the monitoring / message transmission function 6 in the core of the PM in which the resource management space 4 exists detects the crash of the resource use space 2 or the crash of the PM 1 in which the resource use space 2 exists, the resource-use space is used. Identifier correspondence table 5
The resource management space 4 retrieves the destination used by the resource usage space 2 that no longer exists, attaches the destination to the collection notification message, and notifies the resource management space 4 of the destination. Usage space 2 exists
Resources can be collected without being aware that PM1 has crashed.

〔The invention's effect〕

As described above, according to the present invention, even if the resource use space 2 or the PM 1 in which the resource use space 2 crashes, the monitoring / message transmission function 6 in the nucleus of the resource management space 4 substitutes the collection notification message for the resource notification space. Since the configuration of notifying the management space 4 and collecting the resources is adopted, the resources can be collected by the unified mechanism at the time of the resource use space / PM crash. As a result, the crash monitoring by the resource management space 4 becomes unnecessary, and a situation in which unnecessary resources are not collected can be eliminated.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of one embodiment of the present invention, FIG. 2 is an explanatory diagram of resource recovery of the present invention, and FIG. 3 is an explanatory diagram of a conventional technique. In the figure, 1 is a PM (processor module), 2 is a resource use space, 3 is a destination (CAP), 4 is a resource management space, 5 is a resource-use space identifier correspondence table, and 6 is a monitoring / message transmission function.

Claims (1)

(57) [Claims] When a resource is requested from a resource use space, a management resource is created, a resource management space for notifying a destination for specifying the management resource to the resource use space, and use of the allocated resource ends. Then, in a computer system comprising a resource use space for notifying the resource management space of the destination and requesting resource collection, a correspondence table of identifiers of the resource use spaces requesting the resources together with the destination; If the resource use space in which the abnormality has occurred is included in the correspondence table, the corresponding destination is notified to the resource management space on behalf of the resource use space and a request for resource collection is made. And a monitoring / message transmitting means.