TWI635720B - Control method of server system - Google Patents

Abstract

A method for controlling a server system, comprising: a host server sending a first reset signal to a first server, the first server sending a second reset signal to a second server, and so on, Until an (n-1)th server sends an nth reset signal to an nth server. The nth server executes the system startup of the nth server, and returns a first notification signal to the (n-1)th server after completion; the (n-1)th server receives the first notification After the signal, the system of the (n-1)th server is started, and after returning, a second notification signal is sent back to the (n-2)th server, and so on, until the first server receives a message. After the (n-1)th notification signal, the system executing the first server is started, and after the completion, an nth notification signal is sent back to the host server.

Description

Server system control method

The invention relates to a control method of a server system, in particular to a control method of a server system suitable for a multi-stage serial connection structure.

In the prior art, the host server can be allowed to connect to a plurality of server systems in which the external servers are serially connected to achieve a chain structure. For example, systems with non-blade servers often use this chain structure. Figure 1 is a schematic illustration of a prior art servo system 100 of chain structure. The host server 110 can be connected to the external servers 120a to 120d, wherein the external servers 120a to 120d are sequentially connected in a chain structure.

The host server 110 can transmit the reset signals Sa to Sd step by step through the interface cards of the external servers 120a-120d, so that the external servers 120a-120d can sequentially enter the operating state from the standby state, for example, externally. The functional unit in the server is activated. In this mode of operation, since the reset signals Sa to Sd are forwarded in the series, the external server (such as 120d) of the last stage will finally receive the aforementioned reset signal (such as Sd). Therefore, it may happen that the last-stage external server 120d has not received the reset signal Sd, and has not yet started its functional unit, but the detection time of the host server 110 has reached, so the host server 110 has been started, thereby causing the host The server 110 and the external server cannot be properly connected.

When the host server detects the existence of the external server first, the host server does not detect the external server when the connection is made. The reset signal may be sent incorrectly, and the external server does not start successfully. This will cause the host server to fail to boot the server system properly.

Therefore, there is a need in the art for a general solution to improve the operational problems of a chain-like server system.

An embodiment of the present invention provides a method for controlling a server system, including: a host server sends a first reset signal to a first server; and an ith server is provided by an (i-1) server. After receiving an ith reset signal, the i-th server sends an (i+1)th reset signal to an (i+1)th server; when an nth server receives an nth reset After the signal, performing a system startup operation of the nth server; after completing the system startup operation of the nth server, the nth server returns a first notification signal to the (n-1)th servo After a (nk) server receives a kth notification signal from an (n-k+1)th server, performing a system startup operation of the (nk) server; Nk) After the system of the server is started, the (nk) server sends a (k+1) notification signal to a (nk-1) server; when the first server is from a second servo After receiving an (n-1)th notification signal, the system performs a system startup operation of the first server; and after completing the system startup operation of the first server, the first server sends a nth notification News Number to the host server; where 1<i<n, 0<k<(n-1), and i, k, and n are positive integers. The invention relates to a method for operating a server system, which can effectively solve the defects of the chain multi-stage serial server control in the current Internet of Things architecture.

2 and 3 are schematic views of a server system 200 in accordance with an embodiment of the present invention. The server system 200 is shown here in two figures for ease of illustration of the control methods described below. The server system 200 can include a host server 210, and external servers 2201 to 220n, a total of n externally connected servers. The server system 200 can be a chain structure that is connected in series. The server number shown here is, for example, the server 2201 refers to the external first server, that is, the external server closest to the host server 210, the 2202 refers to the external second server, and so on. . This representation is similar in the following, and will not be further described.

4 is a flow chart of a control method 400 of the server system 200 in an embodiment of the present invention. Control method 400 can include:

Step 402: The host server 210 can send the first reset signal R1 to the server 2201;

Step 404: After the server 220i receives the ith reset signal Ri from the server 200(i-1), the server 220i may send the (i+1)th reset signal R(i+1) to the server 220. (i+1);

Step 406: After the server 220n receives the nth reset signal Rn, the system startup operation of the server 220n may be performed;

Step 408: After completing the system startup operation of the server 220n, the server 220n may return the first notification signal A1 to the server 220 (n-1);

Step 410: After the server 220 (n-k) receives the kth notification signal Ak from the server 220 (n-k+1), the system startup operation of the server 220 (n-k) may be performed;

Step 412: After completing the system startup operation of the server 220 (n-k), the server 220 (n-k) may send the (k+1) notification signal A(k+1) to the server 220 (n-k-1);

Step 414: After the server 2201 receives the (n-1)th notification signal A(n-1) from the server 2202, the system startup operation of the server 2201 may be performed;

Step 416: After completing the system startup operation of the server 2201, the server 2201 may send the nth notification signal An to the host server 210;

Step 418: Does the host server 210 receive the nth notification signal An within a predetermined time period? If yes, go to step 422; if no, go to step 420;

Step 420: The first reset signal R1 can be resent to the server 2201, and the process proceeds to step 404;

Step 422: The host server 210 can control the server 2201 to the server 220n.

Where 1<i<n, 0<k<(n-1), and i, k and n are positive integers. The above step 404 is to describe the operation of the server 2202 to the server 220 (n-1) to transmit the reset signal, and the above steps 410 and 412 are to describe the return of the notification signal from the server 220 (n-1) to the server 2202. Operation. The above steps 402 to 406 can refer to FIG. 2, and steps 408 to 416 can refer to FIG.

In step 404, by adjusting the value of the variable i, it can be expressed that the second reset signal R2 is sent from the server 2201 to the server 2202 step by step, and the third reset signal R3 is sent from the server 2202 to the server 2203. The nth reset signal Rn is sent from the server 220 (n-1) to the server 220n. The server 220i receives the (i+1)th reset signal R(i+1) correspondingly after receiving the ith reset signal Ri. In step 406, since the server 220n is the last stage of the chain structure, after receiving the reset signal Rn, the system startup operation of the server 220n can be performed. Details on the system startup operation will be described later. Steps 408 to 414 describe a server in the server 220n to the server 2202, and after receiving the notification signal sent back by the server of the latter stage, perform a system startup operation, and after completing the system startup operation, return Send the notification signal to the previous level server. For example, after the server 2205 completes the system startup operation of the server 2205, the server 2205 can forward the notification signal A(n-4) to the server 2204 to notify the server 2204 to execute the system of the server 2204. Start operation, the notification signal A(n-4) mentioned in this example, the calculation method can refer to the above step 312, the formula can be: nk=5, so (k+1) = (n-5+1 ) = (n-4), then substituting. In other words, when the server performs the system startup operation, it will notify the first-level server to perform the system startup operation before the host server is closer, and so on. Therefore, when the system startup operation of the server 2201 is completed, the server 2201 can send the notification signal An to the host server 210. In step 418, if the host server 210 can receive the notification signal An within the predetermined time period, the host server 210 can confirm that the external servers 2201 to 220n have completed the system startup operation, and can proceed to step 422 by the host server. The servers 2201 to 220n are controlled to control the server system 200. According to an embodiment of the present invention, in step 422, the host server 210 can perform a system startup operation, and control the state of the system startup operation of the servers 2201 to 220n, for example, the release status of the reset pins of the control servers 2201 to 220n, Peripheral Component Interconnect Express (PCIe) interface card status. In step 418, if the host server 210 does not receive the notification signal An within the predetermined time period, it can be determined that at least one of the external servers 2201 to 220n fails to start the system, so the host server 210 can execute Step 420, step 404, and subsequent steps thereof, to re-trigger the system startup operations of the servers 2201 through 220n.

The predetermined time period mentioned in step 418 may be started using a timer in step 402. The predetermined time period may be a longer time period. Therefore, when the notification signal An has not been received yet, the server 2201 may be determined to be more 220n did not successfully perform the system startup operation. According to an embodiment of the invention, the timer may be a watch dog timer, or other suitable timing circuit.

According to the embodiment of the present invention, the system startup operation of the server may be, for example, adjusting the reset pin of the server from the enable level to the disabling level (for example, from 1 to 0, Or vice versa, to cause the server 220n to enter the operational mode from the standby mode. This setting can also be called reset release or reset de-assertion. For example, the standby mode may be that the control unit in the server maintains a standby-by-power supply, but the functional unit in the server is temporarily unavailable; the operation mode may be in the server. The control unit and the functional unit can operate the power supply and activate a set of functional units to enable normal access and operation. The control unit in the server described herein may be, for example but not limited to, a Complex Programmable Logic Device (CPLD), and the functional unit may be, for example but not limited to, a peripheral component A Peripheral Component Interconnect Express (PCIe) unit, a Platform Controller Hub (PCH) unit, and/or a central processing unit (CPU). When a server cannot start the function unit during the system startup operation, it can be re-powered according to the power sequence to restart the function unit by the power supply restart, thereby completing the system startup operation of the server.

According to an embodiment of the present invention, in step 408, when the server 220n completes the system startup operation of the server 220n, the server 220n may receive a reset signal Rn and then pass a predetermined time corresponding to the server 220n. (It may be referred to as the nth predetermined time period), and the server 220n transmits the first notification signal A1 to the server 220 (n-1). The nth predetermined time period may be equal to, or longer than, the length of time required by the server 220n to perform the system startup operation, thereby ensuring that the server 220n has completed its system startup when the server 220(n-1) receives the notification signal A1. operating. According to another embodiment of the present invention, step 408 may include the server 220n receiving the nth reset signal Rn, passing the aforementioned nth predetermined time period, and completing the system startup operation of the server n (for example, all functional units) After being activated and the check signal is generated by the digital circuit control, the server n returns the first notification signal A1 to the server 220 (n-1). This can further ensure that the server 220n has completed its system startup operation.

In the same manner, in step 412, when the server 220 (nk) receives the kth notification signal Ak, it may pass a predetermined time corresponding to the server 220 (nk) (may be called the first (nk) the predetermined time period), the server 220 (nk) returns the (k+1) notification signal A(k+1) to ensure that the server (nk-1) receives the (k+1) notification signal. At A(k+1), the server 220 (nk) has completed its system startup operation. The (n-k)th predetermined period may be set to be longer than the period expected by the system startup operation of the server 220 (n-k). According to another embodiment of the present invention, step 412 may include when the server 220 (nk) receives the kth notification signal Ak, and then passes the (nk) predetermined time period, and after the system startup operation of the server 220 (nk) is completed. (For example, by checking the signal to confirm), the server 220 (nk) sends the return (k+1) notification signal A(k+1) to the server 220 (nk-1). This makes it possible to ensure that the system startup operation of the server 220 (n-k) is indeed completed. The server 220 (n-k) described herein may include the servers 2202 to 220(n-1) shown in FIGS. 2-3.

In the same manner, in step 416, the server may receive the (n-1)th notification signal A(n-1) and then pass the first predetermined time period corresponding to the server 2201. The device 2201 sends the nth notification signal An to the host server 210. According to another embodiment of the present invention, step 416 may be when the server 2201 receives the (n-1)th notification signal A(n-1) and then passes the first predetermined time period corresponding to the server 2201, and the server 2201 The server 2201 transmits the nth notification signal An to the host server 210 after confirming that its system startup operation is completed (for example, a check signal is generated, or a confirmation mode achievable by other circuits).

The nth predetermined time period, the (n-k)th predetermined time period, and the first predetermined time period may be set in the server to be timed.

5 and 6 are schematic views of a server system 200 in accordance with another embodiment of the present invention for use in conjunction with the description of FIG. FIG. 7 is a flow chart of steps that may be further included in the control method 400 of the embodiment of the present invention.

Control method 400 can additionally include the following steps:

Step 710: The host server 210 sends the number of stages calculation signal S1 to the server 2201;

Step 712: After the server 220i receives the series calculation signal Si from the server 220 (i-1), the server 220i sends the series calculation signal S(i+1) to the server 220 (i+1). ;

Step 714: After the server 220n receives the series calculation signal Sn from the server 220 (n-1), the server 220n sends the level reply signal C1 to the (n-1)th server;

Step 716: After the server 220 (nk) receives the level reply signal Ck from the server 220 (n-k+1), it sends a series reply signal C(k+1) to the server 220 (nk- 1);

Step 718: After the server 2201 receives the level reply signal C(n-1) sent from the server 2202, the level reply signal Cn is sent to the host server 210;

Step 720: The server 220 (n-k) generates a time T(n-k) that the server 220 (n-k) should receive the notification signal Ak according to the level reply signal Ck;

Step 722: The server 2201 generates a time T1 when the server 2201 should receive the notification signal A(n-1) according to the series reply signal C(n-1);

Step 724: The host server 210 generates a time Th when the host server 210 should receive the notification signal An according to the level reply signal Cn.

Step 726: The host server 210 has received the notification signal An within the time when the notification signal An should be received. If yes, go to step 422; if no, go to step 402.

Steps 710 to 712 can refer to FIG. 5, and steps 714 to 726 can refer to FIG. Wherein, 1<i<n, 0<k<(n-1), and i, k, and n are positive integers, so the server 220 (nk) described in step 720 may include FIG. 6 and FIG. The servers 2202 to 220 (n-1) are shown.

According to the embodiment of the present invention, the foregoing step 710 may be performed before/after the step 402, or by transmitting a series calculation signal S1 through a path different from the first reset signal R1 (for example, another bus or input/output port). Step 710 is performed in synchronization with step 402. Steps 710 to 726 are used to cause the various stages of the server to confirm the relative number of stages in the chained structure server system.

Figure 8 is a schematic diagram of server system 200 of Figures 5 and 6 at n=5. Figure 8 is for ease of explanation. For example, when there are 5 external servers (corresponding to the above, ie, n=5), the host server 210 can be externally connected to the servers 2201 to 2205, and the number of stages can be sequentially transmitted according to the flow of FIG. The signals S1 to S5 are calculated to the servers 2201 to 2205, and the server 2205 returns the series reply signals C1 to C5 step by step, so that the host server 220 receives the series reply signal C5. The step-by-step number of times of replying signals C1 to C5 can be, for example, a set of numbers, and the step of adding 1 to the step is performed, so that the server 2203 that receives the number of times to reply the signal C2 can be served by the servo. The control unit (e.g., CPLD) in the device 2203 knows that the server 2203 is the third to last in the five-stage external server. Therefore, the control unit in the server 2203 can estimate the reasonable length of time that the server 2203 receives the notification signal A2, which can be the time when the reset signal R4 is sent, plus the time required for the server 2204, 2205 to perform the system startup operation. With a reasonable buffer value, the time T3 at which the server 2203 should receive the notification signal A2 can be obtained, where T3 can be a moment.

According to the embodiment of the present invention, if the server 220 (nk) has not received the kth notification signal Ak at the time when the kth notification signal Ak should be received (which can be expressed as T(nk)), the server 220 (nk) can execute The system of server 220 (nk) initiates operation. For example, in the eighth diagram, if the server 2203 has reached the time T3 but has not received the notification signal A2, it can be inferred that the system startup operation of the server 2204 and/or 2205 is unsuccessful, according to an embodiment of the present invention, Although the server 2203 does not receive the notification signal A2, the system startup operation can be performed by itself, and after the system startup operation of the server 2203 is completed, the notification signal A3 is sent to the server 2202. If the system startup operation of the servers 2202 and 2201 is successful and the signal transmission is successful, the notification signal A5 can be successfully transmitted to the host server 210, so that the host server 210 server can confirm that the external servers 2203, 2202, and 2201 have been successfully executed. The system starts the operation. In this example, the host server 210 may perform subsequent operations using only the servers 2203, 2202, 2201 that have been confirmed to be available, such as performing a platform reset to perform the boot process of the server system, and performing subsequent operations. Data calculations, etc. The remaining servers (such as 2204, 2205) after the server 2203 may not be used in this example.

By using the server system and the control method provided by the embodiment of the present invention, the host server can effectively ensure that each server of the external server serial array has normally performed a system startup operation (for example, warm boot), thereby avoiding the servo. When the system is powered on or other control operations, the functional unit of the external server is not ready. The control method provided by the embodiment of the present invention can be applied to a daisy chain structure or a similarly structured server system, which is helpful for improving the lack of prior art and reducing the failure rate of host server control. The above are only the preferred embodiments of the present invention, and all changes and modifications made to the scope of the present invention should be within the scope of the present invention.

100, 200‧‧‧ server system
110, 210‧‧‧ host server
120a, 120b, 120c, 120d‧‧‧ external server
Sa, Sb, Sc, Sd, R1, R2, R3, R4, R5, R(i-1), Ri, R(i+1), R(n-1), Rn‧‧‧ reset signal
2201, 2202, 220 (i-1), 220i, 220 (i+1), 220 (n-1), 220n, 220 (n-k+1), 220 (nk), 220 (nk-1) ‧ ‧‧server
A1, A2, A3, A4, A5, Ak, A(k+1), A(k+2), A(n-1), An‧‧‧ notification signals
S1, S2, S3, S4, S5, Si, S(i+1), Sn‧‧ ‧ series calculation signals
C1, C2, C3, C4, C5, Ck, C(k+1), C(k+2), C(n-1), Cn‧‧‧ series number confirmation signals
Th, T1, T2, T3, T(nk), T(n-1) ‧ ‧ hours
400‧‧‧Control method
402 to 422, 710 to 720 ‧ ‧ steps

Figure 1 is a schematic illustration of a servo system of a prior art chain structure. 2 and 3 are schematic views of a server system according to an embodiment of the present invention. Fig. 4 is a flow chart showing the control method of the server system of Figs. 2 and 3. 5 and 6 are schematic views of a server system according to another embodiment of the present invention. Fig. 7 is a flow chart showing the steps that can be additionally included in the control method of Fig. 4. Figure 8 is a schematic diagram of the servo system of Figures 5 and 6, at n=5.

Claims (8)

A method for controlling a server system includes: a host server sends a first reset signal to a first server; and an ith server receives an i-th weight from an (i-1) server After the signal is sent, the i-th server sends an (i+1)th reset signal to an (i+1)th server; when an nth server receives an nth reset signal, the first n one of the servers starts the operation; after completing the system startup operation of the nth server, the nth server returns a first notification signal to the (n-1)th server; Nk) after receiving a kth notification signal from an (n-k+1)th server, the server performs a system startup operation of the (nk) server; upon completion of the (nk) server After the system starts the operation, the (nk) server sends a (k+1) notification signal to a (nk-1) server; when the first server receives a message from a second server (n) -1) after the notification signal, performing a system startup operation of the first server; after completing the system startup operation of the first server, the first server sends an nth notification signal to the host server; hair Sending a first-order number calculation signal to the first server; when the i-th server receives one of the i-th level calculation signals transmitted by the (i-1)th server, the i-th server sends An (i+1)th level calculation signal is sent to the (i+1)th server; when the nth server receives the nth series of calculation signals transmitted by the (n-1)th server The nth server sends a first level reply signal to the (n-1)th server; when the (nk) server receives the first (n-k+1) server, After the kth level replies the signal, a (k+1)th level reply signal is sent to the (nk-1)th server; After receiving the (n-1)th level reply signal from the second server, the first server sends an nth level reply signal to the host server; the (nk) server Generating, according to the kth level reply signal, a time when the (k)th server should receive the kth notification signal; the first server generates the first server according to the (n-1)th level reply signal Receiving the time of the (n-1)th notification signal; and the host server generates a time when the host server should receive the nth notification signal according to the nth level reply signal; wherein 1<i<n , 0 < k < (n-1), and i, k, and n are positive integers. The control method of claim 1, further comprising: if the host server receives the nth notification signal within a predetermined time period, the host server controls the first to nth servers. The control method according to claim 1, further comprising: if the host server does not receive the nth notification signal within a predetermined period of time, resending the first reset signal to the first server. The control method of claim 1, wherein one of the system startup operations of each of the servers comprises: if a group of functional units cannot be activated, performing a power restart according to a power sequence. The control method of claim 1, wherein one of the system startup operations of each server comprises: starting a peripheral component interconnection fast (Peripheral Component Interconnect Express; PCle) unit, a Platform Controller Hub (PCH) unit, and/or a central processing unit (CPU). The control method of claim 1, wherein after the system startup operation of the nth server is completed, the nth server returns the first notification signal to the (n-1)th server. After the nth server receives the nth reset signal and then passes the nth predetermined time period, the nth server returns the first notification signal to the (n-1)th server; After the system of the (nk) server starts the operation, the (nk) server sends the (k+1)th notification signal to the (nk-1)th server, and the (nk) server is attached to the (nk) server. Receiving the (k) predetermined period of time after receiving the kth notification signal, the (nk)th server sends the (k+1)th notification signal to the (nk-1)th server; and completing After the system of the first server starts the operation, the first server sends the nth notification signal to the host server, and the first server receives the (n-1)th notification signal. After the first predetermined time period, the first server sends the nth notification signal to the host server. The control method of claim 1, further comprising: if the host server has not received the nth notification signal at the time when the nth notification signal is received, sending another first level calculation signal to the first 1 server. The control method of claim 1, further comprising: if the (nk) server has not received the kth notification signal at the time when the kth notification signal should be received, the (nk) server executes the (nk) The system starts the operation of the server.