JP3084383B2 - Ring communication path failure processing method - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a ring communication path fault processing system, and particularly to a method using a single ring-shaped communication path such as an optical loop between a plurality of processors in a local area network. In a multi-computer system consisting of multiple processors that are connected together, it identifies whether the fault location at the time of a transmission error is on the destination processor side or on the ring communication path, and performs optimal fault handling according to the fault location. The present invention relates to a ring communication path failure processing method capable of selecting the following.

[Conventional technology]

In communication using a ring-shaped communication path such as an optical loop, a transmission processor adds destination address information to transmission data and outputs the data on a ring communication path (hereinafter, referred to as a ring). The output data is transmitted on the ring in a certain direction.

In the receiving processor that is the destination, the communication adapter unit connects each processor to the ring and checks whether the address information of the data matches the destination address. The transmission data is fetched, and a reception response data notification from the receiving side is returned to the transmission source. The reception response data is transmitted in the same direction as the transmission data, and returns to the transmission source after making a round of the ring.

In addition, the physical ring is duplicated, and each communication adapter monitors the disconnection of the ring by using a signal level monitoring unit or the like. In addition, high reliability measures are taken to secure a communication path by reconstructing a normal closed ring by looping back the ring at the adapter unit that detects the ring disconnection.

In the conventional method, if a reception response does not return from the transmission destination within a certain time, data transmission from the transmission source adapter unit was normal, so that the other system did not respond and was uniquely identified as another system failure in the processor. Notifies the transmission requesting program of transmission failure.

[Problems to be solved by the invention]

However, in the conventional method, if a reception response does not return within a certain period of time after data transmission, the transmission from the adapter is assumed to be normal, and the other system is down and does not return a response. In addition to the case where there is a problem, there is also a case where there is a problem on the local system side such as a case where the adapter itself has an error in the communication path looped back by other adapters at both ends of the ring due to a failure in the ring connection part of the sender side adapter For this reason, it is impossible to immediately determine which of the other system and the own system has the true cause of the failure, and there is a problem that the failure processing is delayed.

[Means for solving the problem]

FIG. 1 is a diagram illustrating the principle of the present invention. A plurality of processors are connected to a ring-shaped closed communication path to mutually transmit and receive data, and at the time of a ring communication path failure, loop back a failure site in a communication adapter unit connecting the processor to the ring transmission path, In an inter-processor communication method for reorganizing a normal closed ring communication path and continuing communication, a response monitoring means 1 for monitoring a response from a destination processor after data transmission, and the response monitoring means 1 If it is determined that no response is returned, the retransmission means 3 changes the data transmission destination address from the PCI address of the transmission target processor to the PCI address of the own processor, and retransmits the data to the own processor. Transmission means 3
If the data to the own processor is received normally, the cause of the failure is regarded as another system failure, and if the transmission to the own processor also fails, the transmission failure is sent to the transmission request source program in the processor that transmitted the failure cause as the own system failure. A transmission abnormality notifying unit 5 for notifying, a retransmission test unit 7 for performing a ring return retransmission test addressed to the own processor when data transmission fails, and a failure identification unit 9 for identifying a cause of the failure based on the result obtained by the retransmission test unit 7 If the failure identification means 9 identifies the failure as the own system failure, the communication adapter on the transmitting side is disconnected as a failure, and the transmission is continued using the spare communication adapter. A failure countermeasure unit 11 for notifying that a new transmission request is being made in response to a new transmission request until a restart notification is received from the failure communication adapter unit.

[Action]

The present invention specifies the source adapter itself as the source when the cause of the case where the reception response does not return within a certain time after data transmission is that the communication path looped back by the source adapter is abnormal, By performing ring return data transmission that goes around the network, it is determined whether there is an abnormality cause (other system failure) on the communication destination processor side or an abnormality cause (own system failure) in the source adapter unit including the communication path. .

Therefore, if the response data from the destination does not return within a certain period of time using a ring communication path that can be separated from the working ring and the backup ring, the destination is changed to the source adapter and retransmitted. After identifying whether it is a self-system failure or another-system failure based on the result, the communication abnormality is notified to the transmission request program.

In addition, at the time of the own system failure, the presence or absence of another communication adapter corresponding to the own system processor is checked, and if another communication adapter is present, communication is continued using that communication adapter. On the other hand, in the event of a failure in the other system, an optimum failure process according to the failure location of the own system or the other system is performed by automatically monitoring the recovery of the other system.

〔Example〕

FIG. 2 shows a system configuration diagram of one embodiment of the present invention.
In the figure, a high-speed ring communication path (hereinafter referred to as a ring) 21 includes a working ring 41 and a spare ring 43. Processors 23, 25,
27 and 29 are connected to the ring 21. Communication adapter (PC
I: Processor Communication Interface) 31, 33, 35, 3
7, 39 connect each processor to the ring 21.

FIG. 3 is a block diagram showing functions in each processor according to one embodiment of the present invention. FIG. 2 shows the transmitting processor shown in FIG.
Since 23 is shown as an example, the same parts are denoted by the same reference numerals. The communication target program 301 is a main body for transmitting and receiving data between the processors. The ring transmission / reception control unit 302 exchanges transmission / reception data with the communication target program in accordance with the data transmission / reception request from the communication target program 301. The transmission control unit 303 transmits the transmission data from the ring transmission / reception control unit 302 to the PCI3
Deliver to the ring via one. The reception control unit 304 receives data addressed to the own processor from the ring 21 via the PCI 31.

Further, the retransmission control unit 305, the failure identification control unit 306, and the PCI management unit 307 constitute an abnormality processing unit that controls processing when transmission / reception is abnormal. Retransmission control section 305 performs control to transmit data to ring 21 again when transmission is abnormal. The failure identification control unit 306 is transmitted by the retransmission control unit 305, and performs failure identification based on the result. PCI management unit 307
It has a PCI or spare PCI switching control function.

The ring 21 is composed of two physical rings, a working ring 41 and a spare ring 43. Also, each PCI connected to the ring 21 monitors the health of the ring 21 at the signal level,
Automatically switch from working ring to spare ring when ring 21 breakage is detected
It has a function (ring switching function) to switch to 43 and ensure continuation of communication.

Each means in the above-mentioned claim and each component shown in FIG. 3 are defined as follows. The response monitoring unit 1 is a reception control unit 304, the retransmission unit 3 is a retransmission control unit 305, the transmission abnormality notification unit 5 is a link transmission / reception control unit 302, the retransmission test unit 7 is a transmission control unit 303, and the failure identification unit 9 is failure identification. The control unit 306 and the failure countermeasure unit 11 correspond to the PCI management unit 307, respectively.

FIG. 4 shows an example of ring disconnection due to a source adapter failure. In the case where the disconnection of the ring 21 is detected again even after the PCI switches from the working ring 41 to the spare ring 43, as shown in FIG. 4, the ring 21 is looped back at the PCI at both ends of the disconnection point where the ring disconnection is first detected. And has the function of reconstructing a normal closed ring.

Next, a case where data is transmitted from the processor 23 to the processor 25 as shown in FIG. 2 will be described as an example. In this case, the processor 23 becomes a transmitting processor and the processor 23
25 is a receiving processor.

The transmission processor 23 adds the ring address of the PCI 33, which is the destination address, to the transmission data, and outputs the data to the ring 21 from the working PCI 31.

The output transmission data is transmitted through the working ring 41 in a certain direction.Each PCI checks whether its own ring address matches the transmission destination address, fetches the data addressed to its own PCI, and fetches its own PCI.
To the connected processor 25.

Conversely, the processor 25 that has received the data transmits the reception response notification data to the processor 23. The reception notification data is transmitted on the working ring 41 in the same direction as the transmission data, and is received by the PCI 31.

Next, a process when a transmission / reception error occurs will be described. As an abnormality in which the data transmission / reception processing fails, the failure of the ring 21 can be remedied by the loopback function of each PCI, so that an abnormality in the transmission PCI (own system PCI failure) and an error in the reception processor including the reception PCI failure ( Other system disorders).

For the own system PCI fault, the data transmission can be continued by switching from the working PCI 31 to the spare PCI 39 under the control of the PCI management unit 307.

In case of other system PCI failure, switching to spare PCI has no effect,
It is necessary to wait for the recovery of another system. However, as shown in FIG. 4, even if there is a failure in the connection between the PCI 31 and the ring 21 (indicated by a cross in FIG. 4), it is difficult to isolate the failure from the other system. The response data is detected as an abnormality that does not return to the transmission side within a certain time. When the own system PCI is mistakenly recognized as the other system failure, although the transmission can be continued when the original PCI is switched to the spare PCI 39, a recovery opportunity cannot be permanently obtained.

Next, a method for avoiding the above situation will be described. FIG. 5 shows a flowchart of a process according to one embodiment of the present invention.

In the figure, description is made on the assumption that there is a failure in the ring connection part of the PCI 31 as shown in FIG. Since the data carry signal on the ring 21 is interrupted by PCI31, loopback is performed by PCI33 and PCI39 at both ends of PCI31, and PCI33, PCI3
5, The closed loop of PCI37 and PCI39 is reconfigured. Therefore, at this time, only the PCI 31 is disconnected outside the ring 21.

Now, it is assumed that a transmission request is issued to the processor 25 from the communication target program 301 on the processor 23 shown in FIG. 3 (step 50).

The communication target program 301 requests the ring transmission / reception control unit 302 to transmit by designating transmission data and a communication destination path address of the communication target program 301 of the processor 25 (step 51).

The ring transmission / reception control unit 302 determines the ring address of the PCI 33 of the processor 25 from the path address of the communication destination with the processor 25 that receives the data of the processor 23 (step 5).
2).

The ring transmission / reception control unit 302 manages mapping information between the communication destination path address and the PCI ring address,
The transmission instruction is sent to the transmission control unit 303 using the ring address of I33 as the transmission destination address, and the reception control unit 304 is instructed to monitor the reception response from the PCI 33 (step 53).

The transmission control unit 303 manages the working PCI of the processor 23, adds the ring address of the PCI 33 specified as the transmission destination address (DA) to the transmission data, and outputs the transmission data of the reception response to the working PCI 31. PCI31 is the source address (S
As A), the ring address of the PCI 31 is added to the transmission data, and the data is output onto the ring 21 (step 54).

Since the PCI 31 is now disconnected from the ring 21 due to a failure in the ring connection, the data output in step 54 is not transmitted to the PCI 33. Therefore, the reception control unit 3
In step 04, a response waiting time-out occurs because response data cannot be received within a predetermined time (step 55).

In the conventional method, at this point, no response is received from the other system, and the ring transmission / reception control unit 302 is notified.
From the retransmission control unit 305 provided as an abnormality processing unit.

Hereinafter, a description will be given of a ring return retransmission test, which is a process for accurately separating the own PCI fault from the other system fault, and a recovery method corresponding to the test result.

The retransmission control unit 305 changes the transmission destination address to the address of the own PCI 31 from the data received from the reception control unit 304,
The transmission control unit 303 is again instructed to transmit the same data (step 56).

Along with the processing in step 56, the reception control unit 304 is instructed to monitor the reception response from the own PCI. The PCI 31 transmits data to its own PCI 31 as in the first transmission. If the ring 21 is normal and there is an abnormality in the other system, data is normally received after one round of the ring 21, but in the case of FIG.
Transmission of data addressed to the own PCI 31 also results in a timeout for waiting for a response.

The reception control unit 304 notifies the retransmission control unit 305 of the other system non-response as a reception response monitoring result, as in the case of the first transmission (step 57).

The retransmission control unit 306 sends the second result to the failure identification control unit 306.
Notify. If the second result is normal, the failure identification control unit 306 creates a cause code as the other system failure (step 58), and if the second result is an abnormal system failure (step 59), creates a cause code and sends it to the PCI management unit 307. Notify the own system PCI fault. PCI Management Department
Reference numeral 307 checks the presence or absence of a spare PCI, and if there is a spare PCI, instructs switching of the working PCI. New working PCI
The transmission control unit 303 and the reception control unit 304 are notified of the device address of the PCI 39 to be set (step 60).

Further, the fault identification control unit 306 notifies the ring transmission / reception control unit 302 of the cause code as its own fault (step 6).
1).

The ring transmission / reception control unit 302 instructs the transmission request program to return a transmission error, and notifies the cause code (step 62).

If it is determined that the own PCI fault has occurred at the time of output to the PCI 31 due to a failure in the processor connection unit of the own PCI, the PCI management unit 307
Is notified as a local PCI fault. On the other hand, if the second ring return retransmission succeeds, the other
07, the ring transmission / reception control unit 302 determines that the communication path corresponding to the PCI 33 is in failure, does not accept subsequent transmission requests as in failure of another system, and suppresses unnecessary transmission processing (step 63).

When data is received from the other system PCI corresponding to the communication path during the failure, it is determined that the other system failure has been recovered, and the program using the corresponding communication path is notified of the recovery of the failed path.

If no response is returned within a certain period of time from the time of data transmission, change the destination to the PCI of the transmission source and perform ring return retransmission to disconnect the other system failure or the own system PCI from the ring 21. It correctly identifies whether a ring communication path failure (own system failure) has occurred, and switches to a spare PCI provided separately in the case of a self system failure. In the case of another system failure, processing according to the location of the failure is performed by automatically monitoring the recovery of the other system.

〔The invention's effect〕

As described above, according to the present invention, when the reception response from the transmission destination does not return to the transmission source within a certain period of time after data transmission, the cause of the abnormality is the communication adapter (PCI) of the own system including the ring communication path. Since it is possible to accurately determine whether there is a packet or whether the packet is on the transmission destination processor side, it is useful for improving the reliability of communication processing by performing optimal failure processing for each cause of the failure.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating the principle of the present invention, FIG. 2 is a system configuration diagram of one embodiment of the present invention, FIG. 3 is a block diagram of functions in a processor of one embodiment of the present invention, and FIG. FIG. 5 is a flowchart showing an example of disconnection of a ring due to an original communication adapter failure. FIG. 5 is a flowchart of an embodiment of the present invention. DESCRIPTION OF SYMBOLS 1 ... Response monitoring means, 3 ... Retransmission means, 5 ... Transmission abnormality notification means, 7 ... Retransmission test means, 9 ... Failure identification means, 11 ... Failure recovery means, 21 ... Ring, 23, 25,27,29
... Processor, 31, 33, 35, 37, 39 ... Communication adapter (PC
I), 41: working ring, 43: spare ring, 301: communication target program, 302: ring transmission / reception control unit, 303:
, A transmission control unit, 304, a reception control unit, 305, a retransmission control unit, 306, a failure identification control unit, 307, a PCI management unit.

Continuation of the front page (72) Inventor Tsutomu Aizawa 1015 Uedanaka, Nakahara-ku, Kawasaki City, Kanagawa Prefecture Inside Fujitsu Limited (56) References JP-A-58-124350 (JP, A) JP-A-62-25534 (JP, A) JP-A-2-76337 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) H04L 12/437

Claims (1)

(57) [Claims] 1. A faulty site in a communication adapter section for connecting a plurality of processors to a ring-shaped closed communication path to mutually transmit and receive data and for connecting a processor to a ring transmission path when a ring communication path fails. Response monitoring means for monitoring a response from a destination processor after data transmission, in the inter-processor communication method for reconstructing a normal closed ring communication path and continuing communication by looping back, Retransmission means for changing the data destination address from the PCI address of the target processor to the PCI address of the own processor when it is determined that no response is returned within a certain time, and retransmitting data to the own processor; If the retransmission means normally receives the data addressed to the own processor, the cause of the failure is regarded as another system failure, and the transmission to the own processor also fails. In this case, a transmission error notifying unit for notifying the transmission request source program in the processor that has transmitted the cause of the failure as the own system failure of the transmission failure, a retransmission test unit for performing a ring return retransmission test for the own processor when data transmission fails, Failure identification means for identifying the cause of the failure based on the result obtained by the retransmission test means; and, when the failure identification means identifies the failure as the own system failure, the communication adapter on the transmitting side is disconnected as a failure, and a spare communication adapter is provided. A failure countermeasure unit that continues transmission using the other system, and when identified as another system failure, notifies a new transmission request that another system failure is occurring until a restart notification from the other system failure communication adapter unit is received, A ring communication path failure processing method comprising: