JP4640116B2 - Multi-node computer system, integrated service processor, status management method and program - Google Patents

Description

  The present invention relates to a multi-node computer system in which a plurality of nodes are connected to each other by an inter-node crossbar switch, and more particularly to a status management technique for an inter-node crossbar switch in a multi-node computer system in which inter-node crossbar switches are multiplexed.

  High-speed computers such as supercomputers used in the field of scientific and technological computation are required to have high computing performance. Therefore, in order to satisfy such a requirement, a multi-node computer system in which a plurality of high-performance nodes including a plurality of CPUs and a shared memory shared by them are connected via an inter-node crossbar switch (abbreviated as IXS). Was developed. However, since this multi-node computer system connects the nodes to each other via a single IXS, if an IXS failure occurs, data cannot be exchanged between the nodes, and calculation processing stops. Resulting in.

  In order to solve such a problem, a multi-node computer system in which IXS is multiplexed, that is, a multi-node computer system including a plurality of IXSs that mutually connect nodes is proposed (for example, Patent Document 1). reference).

JP 2000-353154 A

  If IXS is multiplexed as in the multi-node computer system described in Patent Document 1 described above, data can be exchanged via other normal IXS even if a failure occurs in a part of the IXS. Therefore, the calculation process does not stop.

  By the way, in a multi-node computer system in which IXS is multiplexed, each node manages the status of each IXS in order to determine the IXS to be used for data transfer from among the multiplexed IXS. Is required. For example, when a failure occurs in a certain IXS, in order to disconnect the IXS from the multi-node computer system, the status of the IXS managed by each node is changed from the normal state (embedded state) to the degraded state (detached state). When the IXS is restored, in order to incorporate the IXS into the multi-node computer system, it is necessary to change the status of the IXS managed by each node from the degraded state to the normal state.

  In order to perform the status management as described above, for example, the following may be considered.

  In order to change the status of a failed IXS from the normal state to the degraded state, a failure detector is provided in each IXS, and the failure detector that detects the failure of its own IXS sends a failure notification to all nodes. Each node that has received this failure notification changes the status of the IXS where the failure has occurred from the normal state to the degraded state. Further, the failure detector that has detected the failure of the own IXS changes the status of the own IXS managed by the own IXS from the normal state to the degraded state.

  In addition, to change the status of the restored IXS from the degraded state to the normal state, the maintenance personnel manually change the IXS status managed by each node from the degraded state to the normal state, and then recovered. The status managed by IXS is changed from the degraded state to the normal state. Such a status change operation may be performed using, for example, an integrated service processor (abbreviated as an integrated SVP) that manages IXS or a node service processor (abbreviated as a node SVP) that manages nodes. If the integrated SVP and node SVP are used, the status of the recovered IXS (recovered IXS) managed by them is also changed.

  However, the above-described method requires that maintenance personnel manually change the status of the restored IXS, all nodes, the integrated SVP, and the restored IXS managed by all the nodes SVP when IXS is restored. There is a problem that the burden on the worker becomes large, and there is a problem that the status managed by the erroneous operation may be different from the actual status.

  In addition, in order to reduce the burden on the maintenance staff at the time of IXS recovery, the maintenance staff uses the integrated SVP to change the status of the recovery IXS managed by the integrated SVP and all IXS from the degraded state to the normal state, Further, the change contents of the status are notified from the integrated SVP to each node SVP, and each node SVP that has received this notification is notified of the status of the recovery IXS managed by the own node SVP and each node. It is conceivable to change according to According to this method, since the maintenance staff only needs to operate the integrated SVP, the burden on the maintenance staff is reduced.

  However, the above method changes the status when a failure occurs in the node SVP, or when a failure occurs in the communication path between the integrated SVP and the node SVP or between the node SVP and the node. There is a problem that is not performed correctly. The node SVP usually does not affect the operation of the entire multi-node computer system even if a failure occurs. However, in the above method, the node SVP becomes an essential device for the recovery process, and the availability of the entire multi-node computer system is increased. It will decrease.

(Object of invention)
Accordingly, an object of the present invention is to allow maintenance personnel to easily manage the status of the restored IXS managed by each node without using a node service processor when incorporating the IXS recovered from a failure into a multi-node computer system. It is to be able to change to.

A first multi-node computer system according to the present invention includes:
In a multi-node computer system comprising a plurality of nodes, a plurality of inter-node crossbar switches that connect the nodes to each other, and an integrated service processor that manages the inter-node crossbar switches,
The integrated service processor comprises:
When a status change operation request including an identifier of an inter-node crossbar switch to be changed in status and a status after change is input, the status change operation changes the status of the inter-node crossbar switch managed by the own integrated service processor. The configuration changes according to the request, and notifies the status after the change to the inter-node crossbar switch,
The inter-node crossbar switches are respectively
When the changed status is sent from the integrated service processor, the status of the cross-node switch between the own nodes managed by the cross-node switch between the own nodes is changed to the status after the change, and for each node It has a configuration to notify the status after change,
Each of the nodes is
When the changed status is sent from the internode crossbar switch, the status of the internode crossbar switch managed by the own node is changed to the changed status.

A second multi-node computer system according to the present invention includes:
In a multi-node computer system comprising a plurality of nodes, a plurality of inter-node crossbar switches that connect the nodes to each other, and an integrated service processor that manages the inter-node crossbar switches,
The integrated service processor comprises:
A first status storage unit storing the status of the crossbar switch between the nodes;
When a status change operation request including an identifier of an inter-node crossbar switch to be changed in status and a status after change is input by maintenance personnel, it is stored in the first status storage unit according to the status change operation request. First status changing means for changing the status of the corresponding inter-node crossbar switch;
When the status stored in the first status storage unit is changed by the first status change unit, a status change instruction including the changed status is issued to the inter-node crossbar switch whose status has been changed. First status transmitting / receiving means for transmitting,
The inter-node crossbar switches are respectively
A second status storage unit storing the status of the crossbar switch between the own nodes;
Second status change means for changing the contents of the second status storage unit in accordance with a status change instruction sent from the integrated service processor;
Status notification signal transmission means for transmitting a status notification signal indicating the status after the change to each of the nodes when the status stored in the second status storage section is changed by the second status change means; Prepared,
The nodes are respectively
A third status storage unit storing the status of the crossbar switch between the nodes;
And third status change means for changing the contents of the third status storage unit according to a status notification signal sent from the internode crossbar switch.

A third multi-node computer system according to the present invention is the second multi-node computer system,
The inter-node crossbar switches are respectively
When the status change instruction sent from the integrated service processor is an instruction to change to the normal state, the status change instruction includes an initialization hand for initializing the crossbar switch between the own nodes, and
The status notification signal transmitting means is
After the initialization by the initialization means is successful, the status notification signal is transmitted to each node.

A fourth multi-node computer system according to the present invention is the third multi-node computer system,
Each of the nodes is
When the status notification signal sent from the inter-node crossbar switch indicates a normal state, it comprises diagnostic means for diagnosing whether or not the inter-node crossbar switch that is the transmission source of the status notification signal is operable normally ,and,
The third status changing means is
After determining that the inter-node crossbar switch can operate normally by the diagnostic means of all nodes, the status of the inter-node crossbar switch stored in the third status storage unit is changed to a normal state. It has the structure which carries out.

A fifth multi-node computer system according to the present invention is the fourth multi-node computer system,
The second status change means is
When initialization by the initialization unit fails, the status stored in the second status storage unit is changed to a degraded state,
The status notification signal transmitting means is
When the initialization by the initialization unit fails, a status notification signal indicating a degraded state is transmitted to each of the nodes,
The third status changing means is
When a status notification signal indicating a degraded state is sent from the internode crossbar switch, among the statuses stored in the third status storage unit, the internode crossbar switch that is the transmission source of the status notification signal The status is changed to a degraded state.

A sixth multi-node computer system according to the present invention is the fifth multi-node computer system,
The second status change means is
When it is determined that normal operation is impossible by any of the diagnostic means in each node, the status stored in the second status storage unit is changed to a degraded state,
The status notification signal transmitting means is
When it is determined that normal operation is impossible by any of the diagnostic means in each node, a status notification signal indicating a degraded state is transmitted to each node.

A seventh multi-node computer system according to the present invention is the sixth multi-node computer system,
A node service processor for managing each node;
Each of the nodes is
After determining that the inter-node crossbar switch is normally operable by all nodes, a status change instruction is issued to instruct the node service processor to change the status of the inter-node crossbar switch to a normal state. A third status transmission / reception means for transmitting,
The node service processor is
A fourth status storage unit storing the status of the crossbar switch between the nodes managed by each node;
And a fourth status change means for changing the corresponding status to a normal state, which is stored in the fourth status storage unit in accordance with a status change instruction sent from each node.

An eighth multi-node computer system according to the present invention includes:
In a multi-node computer system comprising a plurality of nodes, a plurality of inter-node crossbar switches each connecting the nodes to each other, and an integrated service processor that manages the inter-node crossbar switches,
The integrated service processor comprises:
A first status storage unit storing the status of the crossbar switch between the nodes;
First status change means for changing the contents of the first status storage unit according to a status change instruction sent from the internode crossbar switch;
The inter-node crossbar switches are respectively
A second status storage unit storing the status of the crossbar switch between the own nodes;
Built-in switch operated by maintenance personnel,
Second status changing means for changing the status stored in the second status storage section to a normal state when the built-in switch is operated;
When the status stored in the second status storage unit is changed to the normal state by the second status change unit, the status of the crossbar switch between the own nodes is changed to the normal state for the integrated service processor. Second status transmission / reception means for transmitting a status change instruction for instructing to perform,
A status notification signal for transmitting a status notification signal indicating a normal state to each of the nodes when the status stored in the second status storage unit is changed to a normal state by the second status change unit. A transmission means,
The nodes are respectively
A third status storage unit storing the status of the crossbar switch between the nodes;
A third state in which the status of the inter-node crossbar switch that is the transmission source of the status notification signal is stored in the third status storage unit according to the status notification signal sent from the inter-node crossbar switch is changed to a normal state. The status change means is provided.

A first integrated service processor according to the present invention includes:
An integrated service as a component of a multi-node computer system comprising a plurality of nodes, a plurality of inter-node crossbar switches each interconnecting the nodes, and an integrated service processor for managing the inter-node crossbar switches A processor,
When a status change operation request including an identifier of an inter-node crossbar switch to be changed in status and a status after change is input, the status change operation changes the status of the inter-node crossbar switch managed by the own integrated service processor. The configuration is characterized in that the status is changed according to the request and the status after the change is notified to the inter-node crossbar switch.

The second integrated service processor according to the present invention is the first integrated service processor,
A first status storage unit storing the status of the crossbar switch between the nodes;
When a status change operation request including an identifier of an inter-node crossbar switch to be changed in status and a status after change is input by maintenance personnel, it is stored in the first status storage unit according to the status change operation request. First status changing means for changing the status of the corresponding inter-node crossbar switch;
When the status stored in the first status storage unit is changed by the first status change unit, a status change instruction including the changed status is issued to the inter-node crossbar switch whose status has been changed. And a first status transmission / reception means for transmitting.

The first inter-node crossbar switch according to the present invention is:
Between nodes constituting a component of a multi-node computer system comprising a plurality of nodes, a plurality of inter-node crossbar switches each interconnecting the nodes, and an integrated service processor that manages the inter-node crossbar switches A crossbar switch,
When the changed status is sent from the integrated service processor, the status of the cross-node switch between the own nodes managed by the cross-node switch between the own nodes is changed to the status after the change, and for each node It has the structure which notifies the status after a change, It is characterized by the above-mentioned.

The second internode crossbar switch according to the present invention is the first internode crossbar switch,
A second status storage unit storing the status of the crossbar switch between the own nodes;
Second status change means for changing the contents of the second status storage unit in accordance with a status change instruction sent from the integrated service processor;
Status notification signal transmission means for transmitting a status notification signal indicating the status after the change to each of the nodes when the status stored in the second status storage section is changed by the second status change means; It is characterized by having.

The first node according to the present invention is:
A node that is a component of a multi-node computer system comprising a plurality of nodes, a plurality of inter-node crossbar switches each interconnecting the nodes, and an integrated service processor that manages the inter-node crossbar switches. There,
When the changed status is sent from the internode crossbar switch, the status of the internode crossbar switch managed by the own node is changed to the changed status.

The second node according to the present invention is the following:
A third status storage unit storing the status of the crossbar switch between the nodes;
And third status change means for changing the contents of the third status storage unit according to a status notification signal sent from the internode crossbar switch.

A first status management method according to the present invention includes:
A status management method in a multi-node computer system, comprising: a plurality of nodes; a plurality of inter-node crossbar switches each interconnecting the nodes; and an integrated service processor that manages the inter-node crossbar switches. ,
The integrated service processor comprises:
When a status change operation request including an identifier of an inter-node crossbar switch to be changed in status and a status after change is input, the status change operation changes the status of the inter-node crossbar switch managed by the own integrated service processor. Change according to the request and notify the status after the change to the crossbar switch between the nodes,
The inter-node crossbar switches are respectively
When the changed status is sent from the integrated service processor, the status of the cross-node switch between the own nodes managed by the cross-node switch between the own nodes is changed to the status after the change, and for each node Notify the changed status,
Each of the nodes is
When the changed status is sent from the inter-node crossbar switch, the status of the inter-node crossbar switch managed by the own node is changed to the changed status.

According to a second status management method of the present invention, in the first status management method,
The integrated service processor comprises:
A first status storage unit storing the status of the crossbar switch between the nodes; and
When a status change operation request including an identifier of an inter-node crossbar switch to be changed in status and a status after change is input by maintenance personnel, it is stored in the first status storage unit according to the status change operation request. Processing to change the status of the corresponding cross-node switch between nodes,
A process of transmitting a status change instruction including the changed status to the inter-node crossbar switch whose status has been changed when the status stored in the first status storage unit is changed by the process; Run,
The inter-node crossbar switches are respectively
A second status storage unit storing the status of the crossbar switch between the own nodes; and
A process of changing the contents of the second status storage unit according to a status change instruction sent from the integrated service processor;
When the status stored in the second status storage unit is changed by the process, a process of transmitting a status notification signal indicating the status after the change to each node is executed.
The nodes are respectively
A third status storage unit storing the status of the crossbar switch between the nodes; and
The content of the third status storage unit is changed according to a status notification signal sent from the internode crossbar switch.

The first program according to the present invention is:
An integrated service as a component of a multi-node computer system comprising a plurality of nodes, a plurality of inter-node crossbar switches each interconnecting the nodes, and an integrated service processor for managing the inter-node crossbar switches A program for realizing a processor by a computer including a first status storage unit in which the status of each internode crossbar switch is stored,
The computer,
When a status change operation request including an identifier of an inter-node crossbar switch to be changed in status and a status after change is input by maintenance personnel, it is stored in the first status storage unit according to the status change operation request. First status change means for changing the status of the corresponding inter-node crossbar switch;
When the status stored in the first status storage unit is changed by the first status change unit, a status change instruction including the changed status is issued to the inter-node crossbar switch whose status has been changed. It functions as a first status transmission / reception means for transmission.

The second program according to the present invention is:
Between nodes constituting a component of a multi-node computer system comprising a plurality of nodes, a plurality of inter-node crossbar switches each interconnecting the nodes, and an integrated service processor that manages the inter-node crossbar switches A program for realizing the crossbar switch by a computer having a second status storage unit in which the status of the crossbar switch between the own nodes is stored,
The computer,
Second status change means for changing the contents of the second status storage unit in accordance with a status change instruction sent from the integrated service processor;
When the status stored in the second status storage unit is changed by the second status change unit, the status notification signal transmission unit transmits a status notification signal indicating the changed status to each node. It is characterized by making it.

The third program according to the present invention is:
A node that is a component of a multi-node computer system comprising a plurality of nodes, a plurality of inter-node crossbar switches each interconnecting the nodes, and an integrated service processor that manages the inter-node crossbar switches. , A program to be realized by a computer including a third status storage unit in which the status of each internode crossbar switch is stored,
The computer,
The content of the third status storage unit is made to function as third status change means for changing according to a status notification signal sent from the internode crossbar switch.

[Action]
When a maintenance engineer installs the inter-node crossbar switch (2-0 in FIG. 17) separated from the multi-node computer system in the event of a failure into the multi-node computer system again, the maintenance staff installs the node A status change operation request for requesting that the status of the crossbar switch (2-0 in FIG. 17) be set to the normal state is input.

  In the integrated service processor (1 in FIG. 17) to which the status change operation request is input, the first status change means (11 in FIG. 17) follows the status change operation request in the first status storage unit (FIG. 17). 13) The status of the internode crossbar switch (2-0 in FIG. 17) stored in 13) is changed to the normal state, and the first status transmission / reception means (12 in FIG. 17) further includes the internode crossbar switch. A status change instruction including the changed status is transmitted to (2-0 in FIG. 17).

  In the inter-node crossbar switch (2-0 in FIG. 17) that has received the status change instruction from the integrated service processor (1 in FIG. 17), the second status change means (24 in FIG. 17) stores the second status. The status of the crossbar switch between the nodes (2-0 in FIG. 17) stored in the unit (25 in FIG. 17) is changed to the normal state, and the status notification signal transmission means (27 in FIG. 17) is changed. A status notification signal indicating a later status is transmitted to each node (3-0 to 3-m in FIG. 17).

  In the node (3-0 to 3-m in FIG. 17) from which the status notification signal indicating the normal state is sent from the internode crossbar switch (2-0 in FIG. 17), the third status change means (in FIG. 17). 31) changes the status of the inter-node crossbar switch (2-0 in FIG. 17) stored in the third status storage unit (32 in FIG. 17) to the normal state.

  As described above, a status change operation request for instructing that the maintenance staff change the status of the inter-node crossbar switch (2-0 in FIG. 17) to be incorporated into the multi-node computer system to the normal state is sent to the integrated service processor (FIG. 17). 1), the integrated service processor (1 in FIG. 17), the inter-node crossbar switch (2-0 in FIG. 17) and all nodes (3-0 to 3-m in FIG. 17) Since the status of the corresponding inter-node crossbar switch (2-0 in FIG. 17) can be changed to the normal state, the burden on the maintenance staff can be reduced. In addition, since the above-described processing does not require the node service processor, even if a failure has occurred in the node service processor, the inter-node crossbar switch (2-0 in FIG. 17) is connected to the multi-node computer. Can be incorporated into the system.

  According to the present invention, when an IXS recovered from a failure is incorporated into a multi-node computer system, maintenance personnel can easily change the status of the IXS managed by each node without using a node service processor. Is possible.

  The reason is that, when the status change operation request is input by the maintenance staff, the integrated service processor changes the status of the inter-node crossbar switch managed by the self-integrated service processor according to the status change operation request, and the node The inter-node crossbar switch is configured to notify the status after the change, and when the status after the change is sent from the integrated service processor, the inter-node crossbar switch is managed by the cross-bar switch between the nodes. In addition to changing the status of the crossbar switch between the nodes to the status after the change, each node is notified of the status after the change, and each node sends the status after the change from the crossbar switch between nodes. Managed by the local node when The status of the crossbar switch between nodes because has a configuration to change the status after the change. That is, the maintenance staff uses the node service processor only by inputting a status change operation request for requesting the integrated service processor to change the status of the inter-node crossbar switch incorporated in the multi-node computer system to the normal state. This is because the status of the integrated service processor, the inter-node crossbar switch, and the inter-node crossbar switch managed by all the nodes can be changed to the normal state.

  Next, embodiments of the present invention will be described in detail with reference to the drawings.

[Description of Configuration of First Embodiment]
FIG. 1 is a block diagram showing a configuration example of a first embodiment of a multi-node computer system according to the present invention. Referring to the figure, the multi-node computer system according to the present embodiment is connected to each other via duplexed inter-node crossbar switches (IXS) 2-0 and 2-1 and IXS 2-0 and 2-1. A plurality of connected nodes 3-0 to 3-m, an integrated service processor (integrated SVP) 1 for managing IXS 2-0 and 2-1, and nodes 3-0 to 3-j, ..., 3- (m -J) to 3-m managing node service processors (node SVP) 4-0, ..., 4-k. The integrated SVP1 and the nodes SVP4-0, ..., 4-k are connected to each other by the LAN 5. It is connected. In this embodiment, identifiers “IXS # 0, IXS # 1” are assigned to IXS2-0 and 2-1 respectively, and identifiers “node # 0” are assigned to nodes 3-0 to 3-m, respectively. It is assumed that “node #m” is assigned.

  FIG. 2 is a block diagram illustrating a configuration example of the integrated SVP1, IXS 2-0, node 3-0, and node SVP 4-0.

[Configuration of integrated SVP1]
Referring to FIG. 2, the integrated SVP 1 includes a first status change unit 11, a first status transmission / reception unit 12, a first status storage unit 13, and a console 14.

  In the first status storage unit 13, the statuses of the IXS 2-0 and 2-1 are registered in association with the identifiers. In this embodiment, the status includes a first status indicating whether IXS is an active system (ACT) or a standby system (STB), and a built-in state in which IXS is built in a multi-node computer system (normal). State) or a second status indicating whether it is a disconnected state (degraded state), and three types of states: ACT / normal state, STB / normal state, and STB / degraded state Can be taken.

  The console 14 is an input device for a maintenance person to operate the integrated SVP 1, and the maintenance person can input various operation requests such as a status change operation request using the console 14.

  When the status change operation request including the identifier of the IXS to be changed and the status after the change is input from the console 14, the first status change unit 11 is stored in the first status storage unit 13 accordingly. A function for passing an IXS identifier to be changed to the first status transmission / reception unit 12 and a status after the change, and a status after the change via the first status transmission / reception unit 12 When the status change instruction including the identifier of the transmission source IXS is received, the status stored in the first status storage unit 13 is changed accordingly.

  When the IXS identifier and the changed status are passed from the first status changing means 11, the first status transmission / reception means 12 changes the status including the changed status with respect to the IXS indicated by the identifier. It has a function of transmitting an instruction and a function of passing a status change instruction sent from each IXS 2-0, 2-1 to the first status change means 11.

  The integrated SVP 1 having such functions can be realized by a computer, and when realized by a computer, for example, as follows. A disk, a semiconductor memory, and other recording media recording a program for causing the computer to function as the integrated SVP 1 are prepared, and the computer reads the program. The computer realizes the first status change means 11 and the first status transmission / reception means 12 on its own computer by controlling its own operation according to the read program.

[Configuration of IXS2-0]
The IXS 2-0 includes an initialization unit 21, a failure detection unit 22, a second status transmission / reception unit 23, a second status change unit 24, a second status storage unit 25, and a diagnosis result determination unit 26. And status notification signal transmission means 27. Although not shown in FIG. 2, a data switch for transmitting data sent from a certain node to a designated destination node is also provided in the IXS 2-0. . The other IXSs have the same configuration as IXS2-0.

  The second status change unit 24 has the following functions.

When the status change instruction sent from the integrated SVP 1 is received via the second status transmission / reception means 23, the status of the own IXS 2-0 stored in the second status storage unit 25 in accordance with the received status change instruction The ability to change.
A function of starting the initialization means 21 when the status change instruction is an instruction to change from the degraded state to the normal state.
When the initialization by the initialization unit 21 is successful, the changed status is passed to the status notification signal transmission unit 27, and when the initialization unit 21 fails, the own IXS 2-0 stored in the second status storage unit 25 A function of returning the status to the status before the change and passing the status before the change to the second status transmission / reception means 23.
When the diagnosis result determination means 26 determines that there is an abnormality end indication among the diagnosis results sent from each node, the status of the own IXS 2-0 stored in the second status storage unit 25 is displayed. The function of returning to the status before the change and notifying the second status transmission / reception means 23 and the status notification signal transmission means 27 of the status before the change.

  The second status transmission / reception means 23 has a function of passing the status change instruction sent from the integrated SVP 1 to the second status change means 24, the status passed from the second status change means 24, and the own IXS 2-0. It has a function of transmitting a status change instruction including the identifier “IXS # 0” to the integrated SVP1.

  Only the status of the own IXS 2-0 is stored in the second status storage unit 25.

  When the initialization unit 21 is activated by the second status change unit 24, the initialization unit 21 initializes its own IXS 2-0, or notifies the second status change unit 24 whether the initialization has been successful. Have

  The failure detection means 22 performs the following process when it detects a failure of its own IXS 2-0.

Change the status stored in the second status storage unit 25 in the local IXS 2-0 to “STB / degrade”.
Notify the integrated SVP1 of the occurrence of a failure. Upon receiving this notification, the integrated SVP 1 changes the status of the IXS 2-0 stored in the first status storage unit 13 to “STB / degrade”. Further, if the status of the IXS 2-1 stored in the first status storage unit 13 is “STB / normal”, the status is changed to “ACT / normal”.
-Notify the occurrence of a failure to the other duplexed IXS 2-1. Upon receiving this notification, if the status stored in the status storage unit 25 in its own IXS 2-1 is “STB / normal”, the IXS 2-1 changes it to “ACT / normal”, and thereafter operates. Operates as a system.
Notify each node 3-0 to 3-m that a failure has occurred. Receiving this notification, each of the nodes 3-0 to 3-m changes the status of the IXS 2-0 stored in the third status storage unit 33 in the own node to “STB / degrade”. Further, if the status of the IXS 2-1 stored in the third status storage unit 33 is “STB / normal”, it is changed to “ACT / normal”, and thereafter the IXS 2-1 is recognized as an operational system. Then, data transfer processing between nodes is performed.
Notify each node SVP 4-0 to 4-k that a failure has occurred. Receiving this notification, each of the nodes SVP4-0 to 4-k changes the status of IXS2-0 stored in the fourth status storage unit 43 in its own node SVP to “STB / degrade”. Further, if the status of the IXS 2-1 is “STB / normal”, it is changed to “ACT / normal”.

  The status notification signal transmission unit 27 has a function of transmitting a status notification signal indicating the status of its own IXS 2-0 (the status notified from the status change unit 24) to all the nodes 3-0 to 3-m.

  The diagnostic result judging means 26 has a function for judging whether or not all diagnostic results of diagnostic processing (cable connection check etc.) performed by the nodes 3-0 to 3-m indicate normal termination, A function for notifying each of the nodes 3-0 to 3-m whether or not all of the diagnosis results of .about.3-m indicate normal termination, and one in the diagnosis result of each of the nodes 3-0 to 3-m. However, there is a function of notifying the second status changing means 24 when there is an abnormal end.

  The IXS 2-0 having such a function can be realized by a computer, and when realized by a computer, for example, as follows. A disk, a semiconductor memory, and other recording media on which a program for causing the computer to function as IXS 2-0 is recorded are prepared, and the computer reads the program. The computer controls its own operation according to the read program, so that the initialization unit 21, the failure detection unit 22, the second status transmission / reception unit 23, the second status change unit 24, the diagnosis result determination are performed on the computer. Means 26 and status notification signal transmission means 27 are realized.

[Configuration of node 3-0]
The node 3-0 includes a third status change unit 31, a third status transmission / reception unit 32, a third status storage unit 33, a diagnosis unit 34, and a status notification signal reception unit 35. Although not shown in FIG. 2, the node 3-0 includes a plurality of CPUs and a shared memory shared by the CPUs. The other nodes have the same configuration as that of the node 3-0.

  In the third status storage unit 33, the statuses of the IXS 2-0 and 2-1 are registered in association with the identifiers IXS # 0 and IXS # 1.

  When the third status change means 31 receives the status change instruction via the third status transmission / reception means 32, the third status change means 31 changes the status stored in the third status storage section 33 in accordance with the status change instruction. When the IXS identifier whose status is to be changed and the status after the change are notified from the diagnosis unit 34, the status stored in the third status storage unit 33 is changed, and the third status transmission / reception unit 32 is further updated. In contrast, it has a function of notifying the IXS identifier and the changed status.

  The third status transmission / reception means 32 has a function of passing a status change instruction sent from the node SVP4-0 to the third status change means 31, and an IXS identifier and change notified from the third status change means 31. It has a function of transmitting a status change instruction including a later status to the node SVP4-0.

  The status notification signal receiving unit 35 has a function of passing the status notification signal transmitted from the IXS 2-0 and 2-1 to the diagnosis unit 34.

  The diagnostic means 34 has the following functions.

A function for determining whether the first status of the status indicated by the status notification signal received via the status notification signal receiving means 35 is “normal” or “degrade”.
A function of notifying the third status changing means 31 of the IXS identifier of the status notification signal and the status when the first status is “degrade”.
A function of executing diagnostic processing for IXS that is the transmission source of the status notification signal and transmitting a diagnostic result to the IXS when the first status is “normal”.
A function of notifying the third status change unit 31 of the identifier and status of the IXS when it is notified from the IXS that the diagnosis has been normally completed at all nodes in response to the transmitted diagnosis result.

  The node 3-0 having the above function can be realized by a computer. In this case, for example, the following is performed. A disk, a semiconductor memory, and other recording media recording a program for causing the computer to function as the node 3-0 are prepared, and the computer reads the program. The computer controls the operation of the computer according to the read program, thereby realizing the third status change unit 31, the third status transmission / reception unit 32, the diagnosis unit 34, and the status notification signal reception unit 35 on the computer. .

[Configuration of node SVP4-0]
The node SVP4-0 includes a fourth status change unit 41, a fourth status transmission / reception unit 42, and a fourth status storage unit 43. The other nodes SVP have the same configuration as the node SVP4-0.

  The fourth status storage unit 43 stores the status of each IXS held by each node 3-0 to 3-m in association with the identifier “node # 0 to node #m”. Yes.

  When the fourth status change unit 41 receives the status change instruction from the nodes 3-0 to 3-m via the status transmission / reception unit 42, the fourth status change unit 41 displays the status stored in the fourth status storage unit 43 accordingly. When a status change operation request is sent from the integrated SVP 1 via the LAN 5 or the function to be changed, the status stored in the fourth status storage unit 43 is changed accordingly, and the fourth status transmission / reception means 42 has a function of passing the identifier of the IXS whose status has been changed and the status after the change.

  The fourth status transmission / reception means 42 has a function of passing a status change instruction sent from each of the nodes 3-0 to 3 -m to the fourth status change means 41 and the fourth status change means 41. It has a function of transmitting a status change instruction including an IXS identifier and a changed status to each of the nodes 3-0 to 3-m.

  The node SVP4-0 having such a function can be realized by a computer. In this case, for example, the following is performed. A disk, a semiconductor memory, and other recording media on which a program for causing the computer to function as the node SVP4-0 is recorded are prepared, and the computer reads the program. The computer realizes the fourth status change means 41 and the fourth status transmission / reception means 42 on the own computer by controlling its own operation according to the read program.

[Description of Operation of First Embodiment]
Next, the operation of the present embodiment will be described in detail.

  Now, one of the duplexed IXS 2-0 and 2-1 (for example, IXS 2-0) is operating as the active system, and the other IXS 2-1 is operating as the standby system. That is, it is assumed that the statuses of IXS 2-0 and 2-1 are “ACT / normal” and “STB / normal”, respectively.

  In this state, when the failure detection means 22 in the IXS 2-0 operating as the active system detects a failure in the IXS 2-0, the status stored in the status storage unit 25 in the IXS 2-0, 2-1. Are changed to “STB / degrade” and “ACT / normal”, respectively, and the status storage units 13 and 33 in the integrated server 1, each node 3-0 to 3-m, and each node SVP4-0 to 4-k. , 43, the statuses of IXS 2-0 and 2-1 are also changed to “STB / degrade” and “ACT / normal”, respectively. That is, the IXS 2-1 that was the standby system is switched to the active system, and thereafter, each of the nodes 3-0 to 3-m performs data transfer using the IXS 2-1.

  After that, when the repair of IXS 2-0 that has become “STB / Degrade” is completed, and IXS 2-0 is incorporated into the multi-node computer system, the maintenance staff gives the status of IXS 2-0 from the console 14 to the integrated SVP1. A status change operation request for changing to “STB / normal” is input (step A1 in FIG. 3). This status change operation request includes the identifier “IXS # 0” of IXS 2-0 and the status “STB / normal” after the change.

  When the status change operation request is input, the status change unit 11 in the integrated SVP 1 changes the status of the IXS 2-0 stored in the status storage unit 13 from “STB / degrade” to “STB / normal” accordingly. 3 (step A2 in FIG. 3, step S61 in FIG. 6), and further, the status transmission / reception means 12 is passed the changed status “STB / Normal” and the identifier “IXS # 0” of the IXS 2-0 whose status has been changed ( Step S62). As a result, the status transmission / reception means 12 transmits a status change instruction including the status “STB / normal” after the change to the IXS 2-0 specified by the identifier “IXS # 0” (FIG. 3). Step A3).

  When the status transmission / reception means 23 in the IXS 2-0 receives the status change instruction from the integrated SVP1, it passes it to the status change means 24 in its own IXS 2-0 (step B1 in FIG. 3).

  As a result, the status changing unit 24 changes the status of its own IXS 2-0 stored in the status storage unit 25 from “STB / degrade” to “STB / normal” in accordance with the status change instruction (step in FIG. 3). B2, step S81 in FIG. 8).

  Thereafter, the status changing unit 24 separates the processing based on whether or not the status change performed in step S81 is to change the first status from “degrade” to “normal” (step S82). That is, the process is divided based on whether or not the status change made this time is for incorporating IXS into the multi-node computer system.

  In this example, since the status is changed from “STB / degrade” to “STB / normal” in step S81, the determination result in step S82 is Yes, and the process in step S83 is performed. If the result is No, the status changing unit 24 passes the changed status to the status notification signal transmitting unit 27 (step S88), and each node 3-0 is used using the status notification signal transmitting unit 27. 3-m is notified that the status of its own IXS 2-0 has become “STB / normal”.

  When the determination result in step S82 is Yes, the status change unit 24 activates the initialization unit 22 and initializes its own IXS 2-0 (step B3 in FIG. 3 and step S83 in FIG. 8). Then, it is determined whether or not the initialization is successful (step B4 in FIG. 3 and step S84 in FIG. 8). If successful, the changed status is passed to the status notification signal transmission means 27, and the status notification signal is transmitted. The status notification signal transmitted from the means 27 to all the nodes 3-0 to 3-m is changed from “STB / degrade” to “STB / normal” (step B5 in FIG. 3, FIG. 8 step S88). On the other hand, if initialization fails (No in step B4 in FIG. 3 and No in step 84 in FIG. 8), status return processing is performed (step S85 in FIG. 8), and status change operation of IXS2-0 is performed. Is disabled, and the entire state in the system is returned to the status before step A1 in FIG. 3 is performed. The status return process will be described later.

  When the diagnosis unit 34 in each of the nodes 3-0 to 3-m recognizes that the status notification signal from the IXS 2-0 received by the status notification signal reception unit 35 has changed (step in FIG. 3). C1) As shown in the flowchart of FIG. 10, it is determined whether or not the status of the status notification signal has changed from “STB / degrade” to “STB / normal” (step S101).

  When the status of the status notification signal from the IXS 2-0 changes from “STB / degrade” to “STB / normal” (Yes in step 101), the diagnostic processing for the corresponding IXS 2-0 is performed (step of FIG. 4). C2, step S102 in FIG. 10), otherwise (step 101: No), the status change means 31 is notified of the changed status and the identifier IXS # 0 of IXS2-0 (step S104 in FIG. 10).

  In this example, the status of the status notification signal from the IXS 2-0 has changed from “STB / Degrade” to “STB / Normal” (Yes in Step 101), so the diagnosis unit 34 performs the diagnosis process for the IXS 2-0. (Step C2 in FIG. 4 and Step 102 in FIG. 10). In this diagnosis processing, the diagnosis means 34 in each of the nodes 3-0 to 3-m diagnoses whether the cable connection between the own node and the IXS 2-0 is correctly performed, and notifies the diagnosis result to the IXS 2-0. To do. Thereafter, the diagnosis unit 34 in each of the nodes 3-0 to 3-m waits for a notification sent from the IXS 2-0 (a notification indicating whether or not all the diagnosis results of each node have been normally terminated).

  On the other hand, the IXS 2-0 on which the diagnosis process is performed by the diagnosis unit 34 in each node 3-0 to 3-m is a diagnosis result sent from the diagnosis unit 34 in each node 3-0 to 3-m. Are all normally terminated, and the result is notified to each of the nodes 3-0 to 3-m (steps B6 and B7 in FIG. 4). And when all the diagnostic results are normal completion (step B7 of FIG. 4 is Yes), the process is complete | finished. On the other hand, if there is at least one diagnostic result indicating abnormal termination (No in step B7 in FIG. 4), status return processing is performed (status return processing will be described later), and status change operation of IXS 2-0 is performed. Invalidate all the states in the multi-node computer system to the status before step A1 in FIG.

  The diagnostic means 34 in each of the nodes 3-0 to 3-m waiting for the notification from the IXS 2-0 incorporates the IXS 2-0 into the multi-node computer system based on the notification sent from the IXS 2-0. Is determined (step C3 in FIG. 4 and step S103 in FIG. 10). In other words, when the diagnosis means 34 in each of the nodes 3-0 to 3-m receives a notification from the IXS 2-0 that “the diagnosis result of each node is all normal completion”, the diagnosis means 34 displays the IXS 2-0 as a multi-node. When it is determined that it can be incorporated into the computer system and a notification that “a diagnosis result indicating abnormal termination exists” is received, it is determined that the computer system cannot be incorporated.

  Then, when it is determined that it cannot be incorporated (No in step C3 in FIG. 4 and No in step S103 in FIG. 10), the diagnostic means 34 in each of the nodes 3-0 to 3-m ends the process. On the other hand, if it is determined that it can be incorporated, the diagnosis unit 34 in each of the nodes 3-0 to 3-m sends the status “STB / normal” and IXS2 after the change to the status change unit 31 in its own node. An identifier IXS # 0 of −0 is passed (step S104 in FIG. 10).

  As a result, the status changing means 31 in each of the nodes 3-0 to 3-m changes the status of the IXS 2-0 stored in the status storage unit 33 in the own node to “STB / normal” (FIG. 4). Step C4 in FIG. 11, step S111 in FIG. 11, and the status transmission / reception means 32 in the own node are passed the changed status “STB / Normal” and the identifier IXS # 0 of IXS2-0 (step 112 in FIG. 11). . As a result, the status transmission / reception means 32 in each of the nodes 3-0 to 3-m gives the changed status “STB / normal” and the identifier IXS # 0 of IXS2-0 to the node SVP that manages the node itself. A status change instruction is transmitted (step C5 in FIG. 4).

  The status transmission / reception means 42 in each of the nodes SVP4-0 to 4-k changes the status including the changed status “STB / Normal” and the identifier IXS # 0 of IXS2-0 from the node managed by the own node SVP. When the instruction is sent, the status change instruction is transferred to the status change means 41 in the own node SVP (step D1 in FIG. 4). Thereby, the status change means 41 in each node SVP4-0 to 4-k is managed in association with the node identifier of the transmission source node among the statuses stored in the status storage unit 43. The status of IXS 2-0 is changed to “STB / normal” (step D2 in FIG. 4, step S121 in FIG. 12).

  As described above, the status changed to all devices (integrated SVP1, corresponding IXS, all nodes, all nodes SVP) in the multi-node computer system simply by the maintenance personnel performing an IXS status change operation on the integrated SVP1. Can be reflected.

  Next, when initialization by the initialization unit 21 fails (No in step B4 in FIG. 3 and No in step S84 in FIG. 8), the diagnosis result determination unit 26 determines the diagnosis results of the nodes 3-0 to 3-m. A status return process that is performed when it is determined that there is a diagnosis result indicating abnormal termination (No in step B7 in FIG. 4) will be described.

  If the initialization by the initialization unit 21 fails (step B4 in FIG. 3 is No and step S84 in FIG. 8 is No), the status changing unit 24 in the IXS 2-0 automatically stores the status change unit 24 in the IXS 2-0. The status of IXS 2-0 is returned from “STB / normal” to “STB / degrade” (step B8 in FIG. 5 and step S85 in FIG. 8). Thereafter, the status changing unit 24 passes the returned status to the status transmitting / receiving unit 23, and further passes the returned status to the status notification signal transmitting unit 27 (steps S86 and S87 in FIG. 8).

  Further, the status change unit 24 in the IXS 2-0 determines that the diagnosis result determination unit 26 determines that there is a diagnosis result indicating abnormal termination among the diagnosis results of the nodes 3-0 to 3-m (FIG. 4). Step B7 is No), and the status of the local IXS 2-0 stored in the status storage unit 25 is returned from "STB / Normal" to "STB / Degrade" (Step B8 in FIG. 5 and Step S91 in FIG. 9). Thereafter, the status changing unit 24 passes the returned status to the status transmitting / receiving unit 23, and further passes the returned status to the status notification signal transmitting unit 27 (steps S92 and S93 in FIG. 9).

  When the status “STB / degrade” returned from the status change means 24 is passed to the status transmission / reception means 23 (step S86 in FIG. 8 or step 92 in FIG. 9), the status and the identifier IXS # of its own IXS 2-0 A status change instruction including 0 is transmitted to the integrated SVP 1 (step B9 in FIG. 5).

  When the status transmission / reception means 12 in the integrated SVP 1 receives the status change instruction from the IXS 2-0, it adds it and passes it to the status change means 11 (step A 4 in FIG. 5). As a result, the status change unit 11 returns the status of IXS 2-0 stored in the status storage unit 13 from “STB / normal” to “STB / degrade” (step A5 in FIG. 5, step S71 in FIG. 7). .

  Further, when the status “STB / degrade” returned from the status changing means 24 is passed to the status notification signal transmitting means 27 in the IXS 2-0 (step S87 in FIG. 8, step S93 in FIG. 9), each node 3 The state of the status notification signal transmitted to −0 to 3-m is returned from “STB / normal” to “STB / degrade” (step B10 in FIG. 5).

  When the diagnosis unit 34 in each of the nodes 3-0 to 3-m recognizes that the status notification signal from the IXS 2-0 received by the status notification signal reception unit 35 has changed (step in FIG. 5). C6) As shown in the flowchart of FIG. 10, it is determined whether or not the status of the status notification signal has changed from “STB / degrade” to “STB / normal” (step S101). In the case of this example, the status notification signal state has changed from “STB / Normal” to “STB / Degrade”, so the determination result in Step S101 is No. Therefore, the diagnosis unit 34 passes the returned status “STB / degrade” and the identifier IXS # 0 of the transmission source of the status notification signal to the status change unit 31 in the own node (step 104 in FIG. 10).

  As a result, the status changing unit 31 returns the status of the IXS 2-0 stored in the status storage unit 33 in the own node from “STB / normal” to “STB / degrade” (step C7 in FIG. 5, FIG. 11). Step S111). Thereafter, the status changing unit 31 passes the returned status “STB / degrade” and the identifier IXS of IXS 2-0 to the status transmitting / receiving unit 32 (step S112 in FIG. 11).

  As a result, the status change unit 31 transmits a status change instruction including the returned status “STB / degrade” and the identifier IXS # 0 of IXS2-0 to the node SVP that manages its own node ( Step C8 in FIG.

  The status transmission / reception means 42 in each of the nodes SVP4-0 to 4-k receives a status change instruction including the returned status “STB / degrade” and the identifier IXS # 0 of IXS2-0 from the node managed by the own node SVP. Is sent to the status change means 41 in its own node SVP, the status change instruction and the node identifier of the source node (step D3 in FIG. 5). As a result, the status change means 41 in each of the nodes SVP4-0 to 4-k is managed in association with the node identifier of the transmission source node among the statuses stored in the status storage unit 43. , The status of IXS2-0 is returned to "STB / degrade" (step D4 in FIG. 5, step S121 in FIG. 12).

  With the above operation, the status change operation of IXS 2-0 performed by maintenance personnel is invalidated, and the status of IXS 2-0 managed by each device of the multi-node computer system is the state before performing step A1 in FIG. Can be returned to.

  In the above description, the number of IXSs is two, but any number may be used as long as there are a plurality of IXSs. Further, when the number of IXSs is set to three or more, for example, one of them may be operated as an active system and the rest may be operated as a standby system.

[Effect of the first embodiment]
According to the present embodiment, when an IXS recovered from a failure is incorporated into a multi-node computer system, the status of the recovered IXS managed by each node can be easily maintained without using a node service processor. It becomes possible to change to.

  The reason is that the integrated SVP 1 changes the status of the corresponding IXS (for example, IXS 2-0) stored in the first status storage unit 13 in accordance with the status change operation request input by the maintenance personnel. Means 11 and a first status transmission / reception means 12 for transmitting a status change instruction including the changed status to the corresponding IXS 2-0. The statuses sent from the integrated SVP 1 to each IXS 2-0, 2-1. Second status changing means 24 for changing the status of the local IXS stored in the second status storage section 25 in accordance with the change instruction, and status notification signal transmission for sending a status notification signal indicating the changed status to each node Means 27, and each of the nodes 3-0 to 3-m sends a status notification signal sent from IXS. According because has a third status changing means 31 for changing the status of the IXS stored in the third status storage unit 33. That is, the maintenance staff uses the nodes SVP4-0 to 4-k only by inputting a status change operation request for requesting the integrated SVP1 to set the status of the IXS incorporated in the multi-node computer system to the normal state. This is because the status of the integrated SVP1, the corresponding IXS, and the corresponding IXS managed by all the nodes can be changed to the normal state.

  Further, according to the present embodiment, the status of the IXS managed by each node is changed to the normal state only when the IXS requested to change the status to the normal state can operate normally. Therefore, reliability can be improved.

  The reason is that each of the nodes 3-1 to 3-m includes diagnostic means 34 for diagnosing whether or not an IXS (for example, IXS 2-0) that changes the status to a normal state can operate normally. After the diagnosis means 34 of 1 to 3 -m diagnoses that the IXS 2-0 can operate normally, the status of the IXS 2-0 stored in the third status storage unit 33 in the own node is changed to the normal state. It is because it tries to do.

[Second Embodiment of the Present Invention]
Next, a second embodiment of the multi-node computer system according to the present invention will be described. In the first embodiment described above, when the repaired IXS is re-installed in the multi-node computer system, the installation process starts when the maintenance staff uses the integrated SVP1 console 14 to change the status of the relevant IXS. is doing. For this reason, it is necessary for maintenance personnel to select the IXS from many components on the integrated SVP 1 and perform a status change operation, and it is impossible to deny the possibility that an inexperienced maintenance personnel will make a maintenance error. Further, in the first embodiment, since the IXS installation operation is performed from the integrated SVP1, there is a problem that the IXS installation operation cannot be performed if the integrated SVP1 has a failure and is down. is there. The present embodiment is characterized in that each IXS is provided with a built-in switch that is operated by maintenance personnel when it is incorporated into a multi-node computer system, so that the above-mentioned problems can be dealt with.

[Description of Configuration of Second Embodiment]
FIG. 13 is a block diagram showing an example of the configuration of the second embodiment of the multi-node computer system according to the present invention. The difference from the first embodiment shown in FIG. It is a point provided with IXS2-0a and 2-1a instead of -1.

  FIG. 14 is a block diagram illustrating a configuration example of the integrated SVP1, IXS2-0a, node 3-0, and node SVP4-0 illustrated in FIG. 13, except for the IXS2-0a, the integrated SVP1, Since the configuration is the same as that of the node 3-0 and the node SVP4-0, only the configuration of the IXS 2-0a will be described here. Note that IXS2-1a has the same configuration as IXS2-0a.

  The difference between IXS 2-0a shown in FIG. 14 used in this embodiment and IXS 2-0 shown in FIG. 2 used in the first embodiment is that an embedded switch 28 is added, In addition, a status change unit 24 a is provided instead of the status change unit 24.

  The built-in switch 28 is a mechanical switch provided in the IXS 2-0a. When the IXS 2-0a separated from the multi-node computer system due to the occurrence of a failure is installed again in the multi-node computer system after the repair is completed, Is pressed. When the built-in switch 28 is pressed, processing for incorporating IXS 2-0a into the multi-node computer system is started.

  In addition to the functions provided in the status change unit 24 shown in FIG. 2, the status change unit 24a has its own IXS2- stored in the status storage unit 25 in its own IXS2-0a when the built-in switch 28 is pressed. The status of 0a is changed to “STB / normal”, and the initialization means 21 is activated.

  The IXS 2-0a used in the present embodiment can also be realized by a computer, and when realized by a computer, for example, as follows. A disk, a semiconductor memory, and other recording media on which a program for causing the computer to function as IXS 2-0a is recorded are prepared, and the program is read by the computer. The computer controls its own operation according to the read program, so that the initialization unit 21, the failure detection unit 22, the status transmission / reception unit 23, the status change unit 24a, the diagnosis result determination unit 26, and the status notification are performed on the computer. The signal transmission means 27 is realized.

[Description of Operation of Second Embodiment]
Next, the operation of the present embodiment will be described. Since the operation other than the operation when the maintenance person presses the built-in switch 28 is the same as the first embodiment, only the operation when the maintenance person operates the built-in switch 28 will be described.

  Now, for example, in order to complete the repair of the IXS 2-0a in which a failure has occurred and the status becomes “STB / degrade” and the IXS 2-0a is incorporated into the multi-node computer system again, the maintenance personnel Assume that the built-in switch 28 is pressed (step E1 in FIG. 15).

  When the built-in switch 28 is pressed, the status changing unit 24a in the IXS 2-0a changes the status of the local IXS 2-0a stored in the status storage unit 25 from “STB / degrade” to “STB / normal”. (Step E2 in FIG. 15 and Step S161 in FIG. 16). Thereafter, the status changing unit 24a passes the changed status “STB / normal” to the status transmitting / receiving unit 23 (step S162 in FIG. 16). Thereby, the status transmission / reception means 23 transmits a status change instruction including the changed status “STB / normal” and the identifier IXS # 0 of the own IXS 2-0a to the integrated SVP 1 (step E3 in FIG. 15).

  When the status transmission / reception means 12 in the integrated SVP 1 receives the status change instruction from the IXS 2-0a, it passes it to the status change means 11 (step E4 in FIG. 15). As a result, the status changing unit 11 changes the status of the IXS 2-0a stored in the status storage unit 13 from “STB / degrade” to “STB / normal” (step E5 in FIG. 15, step S71 in FIG. 7). ).

  On the other hand, when the processing in step 162 is completed, the status changing unit 24a in the IXS 2-0a activates the initializing unit 21 to initialize the IXS 2-0a (step B3 in FIG. 15 and step S163 in FIG. 16). .

  If initialization is successful (step B4 in FIG. 15 is Yes and step S164 in FIG. 16 is Yes), the same processing as Step B5 and Step S88 described above (Step B5 in FIG. 15 and Step S168 in FIG. 16). ) Is performed. As a result, the status of IXS 2-0a is also reflected in each of the nodes 3-0 to 3-m and each of the nodes SVP4-0 to 4-k.

  On the other hand, if initialization fails, the same processing as in steps B8 to B10 and steps S85 to S87 described above (steps B8 to B10 in FIG. 5 and steps S165 to S167 in FIG. 16) is performed. The status of 0a is returned to “STB / degrade”.

[Effects of Second Embodiment]
According to the present embodiment, in addition to the effects of the first embodiment, when an IXS that has been repaired by a maintenance staff is incorporated into a multi-node computer system, the installation operation is mistaken or forgotten. The effect that operation mistakes can be prevented can be obtained. Furthermore, even when a failure occurs in the integrated SVP1, it is possible to obtain an effect that IXS can be incorporated into a multi-node computer system.

  The reason is that each IXS 2-0, 2-1 has a built-in switch 28, and when the built-in switch 28 is pressed, processing for incorporating IXS into the multi-node computer system is started. Because.

  The present invention can be applied to a spar computer or the like in which a plurality of nodes are connected to each other via an inter-node crossbar switch.

1 is a block diagram illustrating a configuration example of a first embodiment of a multi-node computer system according to the present invention. It is a block diagram which shows the structural example of integrated SVP1, IXS2-0, the node 3-0, and the node SVP4-0. It is a flowchart for demonstrating operation | movement of 1st Embodiment. It is a flowchart for demonstrating operation | movement of 1st Embodiment. It is a flowchart for demonstrating operation | movement of 1st Embodiment. It is a flowchart which shows the process example of the 1st status change means 11 when a status change operation request | requirement is input. It is a flowchart which shows the process example of the 1st status change means 11 when a status change instruction | indication is received. It is a flowchart which shows the process example of the 2nd status change means 24 when a status change instruction | indication is received. It is a flowchart which shows an example of the process which the 2nd status change means 24 performs when there exists a diagnostic result which shows abnormal termination in the diagnostic result sent from each node. 4 is a flowchart illustrating a processing example of a diagnosis unit 34. It is a flowchart which shows the process example of the 3rd status change means 31 when a status is notified from the diagnostic means. It is a flowchart which shows the process example of the 4th status change means 41 when a status change instruction | indication is received. It is a block diagram which shows the structural example of 2nd Embodiment of the multi-node computer system concerning this invention. It is a block diagram which shows the structural example of integrated SVP1, IXS2-0a, the node 3-0, and the node SVP4-0. It is a flowchart for demonstrating operation | movement of 2nd Embodiment. It is a flowchart which shows the process example of the 2nd status change means 24a when the built-in switch 28 is pressed down. It is a figure for demonstrating the effect | action of this invention.

Explanation of symbols

1 ... Integrated service processor (integrated SVP)
DESCRIPTION OF SYMBOLS 11 ... 1st status change means 12 ... 1st status transmission / reception means 13 ... 1st status storage part 14 ... Console 2-0-2-n, 2-0a, 2-1a ... Crossbar switch between nodes (IXS)
21 ... Initialization means 22 ... Fault detection means 23 ... Second status transmission / reception means 24, 24a ... Second status change means 25 ... Second status storage unit 26 ... Diagnosis result judgment means 27 ... Status notification signal transmission means 28 ... built-in switches 3-0 to 3-m ... node 31 ... third status change means 32 ... third status transmission / reception means 33 ... third status storage section 34 ... diagnosis means 35 ... status notification signal reception means 4-0 ~ 4-k ... Node service processor (node SVP)
41 ... Fourth status change means 42 ... Fourth status transmission / reception means 43 ... Fourth status storage 5 ... LAN

Claims (5)

In a multi-node computer system comprising a plurality of nodes, a plurality of inter-node crossbar switches that connect the nodes to each other, and an integrated service processor that manages the inter-node crossbar switches,
The integrated service processor comprises:
A first status storage unit storing the status of the crossbar switch between the nodes;
When a status change operation request including an identifier of an inter-node crossbar switch to be changed in status and a status after change is input by maintenance personnel, it is stored in the first status storage unit according to the status change operation request. First status changing means for changing the status of the corresponding inter-node crossbar switch;
When the status stored in the first status storage unit is changed by the first status change unit, a status change instruction including the changed status is issued to the inter-node crossbar switch whose status has been changed. First status transmitting / receiving means for transmitting,
The inter-node crossbar switches are respectively
A second status storage unit storing the status of the crossbar switch between the own nodes;
Second status change means for changing the contents of the second status storage unit in accordance with a status change instruction sent from the integrated service processor;
Status notification signal transmitting means for transmitting a status notification signal indicating the changed status to each of the nodes when the status stored in the second status storage section is changed by the second status change means;
Status change instruction transmitted from the integrated service processor, when is an indication of the change to the normal state, and a initial catheter stage for initializing the crossbar switch between its own node,
The nodes are respectively
A third status storage unit storing the status of the crossbar switch between the nodes;
Third status change means for changing the contents of the third status storage unit according to a status notification signal sent from the inter-node crossbar switch;
When the status notification signal sent from the inter-node crossbar switch indicates a normal state, it is a diagnostic process for the inter-node crossbar switch that is the transmission source of the status notification signal, and is a connection with the own node A diagnostic means for performing a diagnostic process including a check , and
The status notification signal transmission unit has a configuration for transmitting the status notification signal to each of the nodes after the initialization by the initialization unit is successful.
When the diagnosis result of the diagnosis means of all the nodes is normal completion, the third status change means sets the status of the inter-node crossbar switch stored in the third status storage unit to a normal state. A multi-node computer system having a configuration for changing. The multi-node computer system of claim 1 .
The second status change means is
When initialization by the initialization unit fails, the status stored in the second status storage unit is changed to a degraded state,
The status notification signal transmitting means is
When the initialization by the initialization unit fails, a status notification signal indicating a degraded state is transmitted to each of the nodes,
The third status changing means is
When a status notification signal indicating a degraded state is sent from the internode crossbar switch, among the statuses stored in the third status storage unit, the internode crossbar switch that is the transmission source of the status notification signal A multi-node computer system characterized by having a configuration for changing the status of the system to a degraded state. The multi-node computer system according to claim 2 , wherein
The second status change means is
When it is determined that normal operation is impossible by any of the diagnostic means in each node, the status stored in the second status storage unit is changed to a degraded state,
The status notification signal transmitting means is
A multi-node computer system comprising: a status notification signal indicating a degraded state to each node when it is determined that normal operation is impossible by any of the diagnostic means in each node. The multi-node computer system according to claim 3 .
A node service processor for managing each node;
Each of the nodes is
After determining that the inter-node crossbar switch is normally operable by all nodes, a status change instruction is issued to instruct the node service processor to change the status of the inter-node crossbar switch to a normal state. A third status transmission / reception means for transmitting,
The node service processor is
A fourth status storage unit storing the status of the crossbar switch between the nodes managed by each node;
And a fourth status change means for changing the corresponding status to a normal state, which is stored in the fourth status storage unit in accordance with a status change instruction sent from each node. Node computer system. An integrated service processor computer having a first status storage unit, a first status change unit, and a first status transmission / reception unit, respectively, a second status storage unit, a second status change unit, and a status notification A plurality of inter-node crossbar switch computers having signal transmission means and initialization means, and a plurality of node computers each having a third status storage unit, third status change means, and diagnosis means Is a status management method executed by
The inter-node crossbar switch computer connects the node computers to each other,
The first status storage unit stores the status of the inter-node crossbar switch computer,
In each of the second status storage units, the status of the computer for crossbar switch between own nodes is stored,
In each of the third status storage units, the status of each of the inter-node crossbar switch computers is stored.
When a status change operation request including the identifier of the inter-node crossbar switch computer to be changed by the first status change means and the status after the change is input by the maintenance staff, the first status change means performs the change according to the status change operation request. Change the status of the corresponding inter-node crossbar switch stored in the first status storage unit,
When the status stored in the first status storage unit is changed by the first status change unit by the first status transmission / reception unit, the inter-node crossbar switch computer whose status is changed , Send a status change instruction with the changed status,
The second status change unit of the inter-node crossbar switch computer changes the contents of the second status storage unit of its own computer according to the status change instruction sent from the integrated service processor computer,
When the status change instruction sent from the integrated service processor computer instructs the initialization means of each inter-node crossbar switch computer to change to the normal state, Perform initialization,
Status indicating the status stored in the second status storage unit of the local computer after the status notification signal transmission unit of the crossbar switch computer between the nodes has been successfully initialized by the initialization unit of the local computer Sending a notification signal to the computer for each node;
The third status change unit of each of the node computers changes the contents of the third status storage unit of its own computer according to a status notification signal sent from the internode crossbar switch computer,
When the status notification signal sent from the inter-node crossbar switch computer indicates a normal state, the diagnostic means of each node computer is for the inter-node crossbar switch that is the transmission source of the status notification signal A diagnostic process for a computer, including a diagnostic process including a connection check with the own computer,
The third status change unit changes the status of the inter-node crossbar switch stored in the third status storage unit to a normal state when the diagnosis result of the diagnostic unit of all the nodes is the control end. A status management method characterized by:

Images