TWI598728B - Server and method for detecting power reset - Google Patents

Description

Server and its baseboard management controller and judgment server Whether the event is caused by power interruption

The present invention relates to a system and method for detecting errors, and more particularly to a server and a method for determining whether a server reset event is a power interruption.

The existing server receives AC power from an external source. However, if the AC power is interrupted for several reasons, such as the interruption of the AC power caused by the accidental disconnection of the power plug of the server, the operation of the server will be affected, and the system may be shut down or the service may be interrupted.

At present, detecting the power interruption of the server and resetting it, the high-order power supply with the function of detecting the AC power interruption can be used to supply the power required by the server, and the substrate management control of the server operating condition is controlled. The device obtains a message of a power reset event from the power supply, but the disadvantage of this method is that the high-order power supply with the function of detecting the alternating current power interruption needs to be specially purchased, and the use of the high-order power supply will affect the system. Compatibility, therefore, it is the current research direction to detect server power interruption without using a high-order power supply.

Accordingly, it is an object of the present invention to provide a method for determining whether a server reset event is reset by a server that is reset by a firmware of a firmware without using a high-order power supply.

Therefore, the method for determining whether the server reset event is caused by a power interruption is performed by a server, the server includes a baseboard management controller, and the baseboard management controller includes a random access memory and an electric a processing unit connected to the random access memory, and an event recording unit electrically connected to the processing unit, the random access memory stores a power reset flag, and the power reset detection method includes a step (A ), one step (B), one step (C), and one step (E).

The step (A) is that the processing unit determines whether to initialize the random access memory according to the received event signal.

The step (B) is a parameter value of the processing unit according to the power reset flag to determine whether it is a protocol abnormality.

In the step (C), if the result of the determination in the step (B) is NO, the processing unit determines whether it is an own operation abnormality.

The step (E) is that if the result of the determination in the step (C) is NO, the processing unit records a power reset event to the event recording unit.

Furthermore, another object of the present invention is to provide a server.

Thus, the server includes a substrate management controller including a random access memory and a processing unit.

The random access memory stores a power reset flag.

The processing unit is electrically connected to the random access memory, according to the received event signal, to determine whether to initialize the random access memory, and according to a parameter value of the power reset flag, to determine whether it is a protocol abnormality If the determination result is no, the processing unit determines whether it is an own operation abnormality, and if the determination result is also no, the processing unit records a power reset event to the event recording unit.

The effect of the invention is that the server can directly detect the power reset event without purchasing a high-order power supply, thereby not affecting system compatibility.

1‧‧‧Server

11‧‧‧Base management controller

111‧‧‧ Random access memory

112‧‧‧Processing unit

113‧‧‧ Event Recording Unit

114‧‧‧Monitoring unit

115‧‧‧Member layer unit

116‧‧‧Electronic erasing read-only memory

12‧‧‧IPMI interface

13‧‧‧South Bridge

2‧‧‧Remote Server

21‧‧‧IPMI client interface

A~E‧‧‧Steps

Other features and effects of the present invention will be apparent from the embodiments of the present invention, wherein: FIG. 1 is a schematic diagram illustrating a server of the present invention; and FIGS. 2A and 2B are flowcharts illustrating the servo. A first embodiment of a power reset detection method performed by the device; FIG. 3 is a flowchart illustrating a second embodiment of the power reset detection method performed by the server; FIG. 4 is a flowchart illustrating A third embodiment of the power reset detection method performed by the server.

Before the present invention is described in detail, it should be noted that in the following description, similar elements are denoted by the same reference numerals.

Referring to FIG. 1, a server 1 of the present invention includes a substrate tube Baseboard Management Controller (BMC) 11, a South Bridge 13, and an Intelligent Platform Management Interface (IPMI) 12.

The smart platform management interface 12 determines whether to output a reset request signal according to a reset command. The reset command may be from a smart platform management client interface (IPMI client) 21 of a remote server 2, requesting a cold reset, a warm reset, or a firmware update at the remote end. One of the commands is issued; the reset command may also come from the south bridge 13 of the server 1, and is issued when the local end requests one of a cold reset, a warm reset, or a firmware update.

The baseboard management controller 11 includes a random access memory (RAM) that stores a power reset flag, a processing unit 112 that electrically connects the random access memory 111, and an electrical connection unit. And recording an event log 113 of a power reset event, a watchdog 114 electrically connecting and monitoring the operation of the processing unit 112, electrically connecting the processing unit 112, and being the processing unit 112 An agreement layer unit (IPMI stack) 115 of the software stack structure of the operating system, and an electrically-erasable programmable read-only memory (EEPROM) 116 electrically connected to the processing unit 112. In this example, the random access memory 111 is a dynamic random access memory (DRAM).

2A and 2B, the first embodiment of the method for determining whether a server reset event is a power interruption is performed by the servo. The device 1 executes and includes the following steps.

Referring to FIG. 2A, first, in step (A), the processing unit 112 determines whether to initialize the random access memory 111 based on the received event signal. The event signal is an input power signal required for supplying the server 1, a processing abnormality signal from the monitoring unit 114 indicating that the processing unit 112 is abnormal, and a reset request signal from the smart platform management interface 12. And one of the agreed exception signals from the protocol layer unit 115. Moreover, step (A) includes the following sub-steps.

In step (A1), the processing unit 112 receives one of the input power signal, the processing abnormality signal, and the reset request signal, that is, proceeds to step (A2). In detail, this step (A1) includes the following sub-steps.

In step (A11), the processing unit 112 receives the processing abnormality signal and proceeds to step (A2).

In step (A12), the processing unit 112 receives the reset request signal.

In step (A13), the processing unit 112 writes a reset request flag value to the electronic erasure type read only memory 116, and proceeds to step (A2).

In step (A14), the processing unit 112 receives the input power signal and proceeds to step (A2).

In step (A2), the processing unit 112 is restarted, and the random access memory 111 is initialized, and proceeds to step (B).

Step (A3), the processing unit 112 receives the agreement abnormality signal, and determines that the random access memory 111 is not restarted and is not initialized. Go directly to step (B).

In addition, the event signal (the input power signal, the processing abnormal signal, the reset request signal, and the agreement abnormal signal) are respectively caused by the following trigger events: 1. Supplying the server 1 The AC power interruption is restored and the processing unit 112 is reset; 2. The processing unit 112 itself is abnormally operated (downtime) is reset by the monitoring unit 114; 3. The smart platform management client interface from the remote server 2 is used. 21. The reset command from the south bridge 13 causes the processing unit 112 to perform a reset request operation such as a cold reset, a warm reset, or an update firmware reset according to the reset command; 4. The protocol layer unit 115 The occurrence of a contract exception causes the agreement layer unit 115 to reset. Therefore, according to the different triggering events, the processing unit 112 is divided into advanced steps (A2) and then proceeds to step (B) according to the corresponding received event signal, and directly proceeds to step (B) two different Operating mode.

Referring to FIG. 2B, the processing unit 112 must then distinguish which of the received event signals is the one, and in this example, the step (D) performed prior to step (B), the processing unit 112 is based on The reset request flag value of the electronically erased read-only memory 116 is used to determine whether it is a reset request operation, and includes the following sub-steps.

In step (D1), the processing unit 112 compares whether the reset request flag value meets a set value, and if not, the processing unit 112 determines that the reset request operation is not, and performs step (B).

Step (D2), if the result of the comparison of the step (D1) is YES, the processing unit 112 determines that the reset request operation.

In step (D3), the processing unit 112 dumps the reset request flag value to the random access memory 111.

In step (D4), the processing unit 112 sends a clear command to clear the reset request flag value of the electronic erasure type read only memory 116.

As described herein, according to the electrical characteristics of the electronic erasable read-only memory 116, when the processing unit 112 receives the reset request signal, writes the reset request flag value and restarts, and does not The reset request flag value stored in the electronic erasure type read-only memory 116 disappears. Therefore, when the reset request flag value meets the set value, the reset request operation can be determined and executed. The write can be rewritten when the processing unit 112 receives the reset request signal again.

Next, in step (B), the processing unit 112 determines whether it is a contract abnormality according to a parameter value of the power reset flag, and includes the following sub-steps.

In step (B1), the processing unit 112 compares whether the parameter value of the power reset flag meets a preset value.

In the step (B2), if the result of the comparison in the step (B1) is YES, the processing unit 112 determines that the agreement layer unit 115 has a contract abnormality.

Step (B3), if the result of the comparison in step (B1) is no, the processing unit 112 writes the preset value into the power reset flag of the random access memory 111 to update the parameter value, and Perform step (C).

It is explained that, according to the electrical characteristics of the random access memory 111, when the processing unit 112 restarts and initializes the random access memory 111, the power reset stored in the random access memory 111 The parameter value of the flag disappears, but since the processing unit 112 receives the agreement abnormal signal and does not initialize the random access memory 111, when the parameter value of the power reset flag meets the preset value, It is judged that the agreement is abnormal.

Next, in step (C), the processing unit 112 determines whether it is a self-operation abnormality, and includes the following sub-steps.

In step (C1), the processing unit 112 determines whether it is an operation abnormality according to a logic value of the timeout flag indicating that the processing unit operates abnormally by the monitoring unit 114. When the processing unit 112 determines that the overflow time flag is in a second logic, the processing unit 112 determines that the self operation is not abnormal, and determines that the power reset event, and proceeds to step (E).

Step (C2), if the processing unit 112 determines that the overflow time flag is in a first logic, the processing unit 112 determines that the self operation is abnormal.

Next, in step (E), the processing unit 112 records the power reset event to the event recording unit 113.

As can be seen from the above description, in this example, the processing unit 112 sequentially excludes the reset request operation via the step (D), and then excludes the agreement exception by the step (B), and finally excludes the step (C) as the If the operation is abnormal, the server 1 is judged to have undergone power reset, and the power reset event is recorded to the event recording unit 113.

Referring to FIG. 3, a second embodiment of the method for determining whether a server reset event is caused by a power interruption is compared with the first embodiment. It is similar, except that the step (D5) performed after the step (C1) is also included.

In step (D5), if the processing unit 112 determines that the self-operation is abnormal according to the overflow flag in step (C1), the processing unit 112 compares the random access memory 111 with the random access memory 111. Whether the reset request flag value meets the set value, and if so, repeats the step (D2); if not, executes step (E).

Thereby, it can be prevented that after the power reset event occurs in the reset request operation of the step (D2), the reset request flag value stored by the electronic erase type read only memory 116 is not caused by The power is reset and disappears, so that the processing unit 112 does not detect the situation of the power reset event. At this time, by the processing unit 112 performing the transfer of the reset request flag value in step (D3), the reset of the random access memory 111 can be determined by the step (C1). Whether the request flag value disappears due to the power reset to confirm whether the power reset event still occurs after the step (D2).

Referring to FIG. 4, a third embodiment of the method for determining whether a server reset event is a power interruption is a variation of the first and second embodiments, and the difference is that the step (D) is The step (C) is performed after, and the step (D) includes the step (D1'), the step (D2'), and the step (D4').

In step (D1'), the processing unit 112 compares whether the reset request flag value meets a set value, and if not, the processing unit 112 determines the power reset event and performs step (E).

Step (D2'), if the result of the comparison of the step (D1') is YES, the processing unit 112 determines that it is a reset request operation.

In step (D4'), the processing unit 112 sends a clear command to clear the reset request flag value of the electronically erasable read only memory 116.

As can be seen from the above description, in the third embodiment, the processing unit 112 sequentially excludes the agreement exception via the step (B), and then excludes the self operation abnormality by the step (C), and finally excludes the step (D). For the reset request operation, it is judged that the server 1 has undergone power reset, and the power reset event is recorded to the event recording unit 113.

In summary, the method of the present invention for determining whether a server reset event is a power interruption has the following advantages.

First, the high-order power supply is not required, and the determination is only performed by the processing unit 112 of the baseboard management controller 1, that is, whether the power reset event of the server 1 can be recorded.

Secondly, in the detection of the present invention, the processing unit 112 can exclude any one of the reset request operation, the self-operation abnormality, or the agreement abnormality, and finally correctly determine that the server 1 passes the power. The reset is indeed achieved for the purpose of the present invention.

The above is only the preferred embodiment of the present invention, and the scope of the present invention is not limited thereto, that is, the simple equivalent changes and modifications made by the patent application scope and patent specification content of the present invention, All remain within the scope of the invention patent.

1‧‧‧Server

11‧‧‧Base management controller

111‧‧‧ Random access memory

112‧‧‧Processing unit

113‧‧‧ Event Recording Unit

114‧‧‧Monitoring unit

115‧‧‧Member layer unit

116‧‧‧Electronic erasing read-only memory

12‧‧‧IPMI interface

13‧‧‧South Bridge

2‧‧‧Remote Server

21‧‧‧IPMI client interface

Claims (10)

A method for determining whether a server reset event is caused by a power interruption is performed by a server, the server includes a baseboard management controller, the baseboard management controller includes a random access memory, and an electrical connection to the random a processing unit that accesses the memory, the random access memory stores a power reset flag, and the power reset detection method includes the following steps: (A) the processing unit determines according to the received event signal Whether to initialize the random access memory; (B) the processing unit determines whether it is a protocol abnormality according to a parameter value of the power reset flag; (C) if the determination result of the step (B) is negative, the The processing unit determines whether it is an own operation abnormality; and (E) if the determination result of the step (C) is negative, the processing unit determines that the power reset event belongs to the server, and the power reset event is about the supply A server reset caused by a power interruption to the server. The method of claim 1, wherein the determining whether the server reset event is caused by a power interruption, the baseboard management controller further includes: a monitoring unit electrically connecting and monitoring the operation of the processing unit; and an electrical connection to the processing unit a protocol layer unit, the event signal being one of an input power signal, a processing exception signal from the monitoring unit indicating that the processing unit is operating abnormally, a reset request signal, and a protocol abnormality signal from the protocol layer unit, Wherein step (A) comprises: (A1) the processing unit receives one of the input power signal, the processing abnormality signal, and the reset request signal, that is, proceeds to step (A2), (A2) the processing unit restarts, and initializes the random storage The memory is taken, and the process proceeds to step (B). (A3) The processing unit receives the agreement abnormality signal, and determines that the random access memory is not restarted and does not initialize, and proceeds directly to step (B). The method for determining whether the server reset event is a power interruption as described in claim 2, and the step (B) comprises: (B1) the processing unit comparing whether the parameter value of the power reset flag conforms to a pre- If the result of the comparison in step (B1) is YES, the processing unit determines that the agreement layer unit has a contract exception, and (B3) if the result of the step (B1) comparison is no, the processing unit will The preset value is written to the power reset flag of the random access memory to update the parameter value. The method of determining whether the server reset event is a power interruption as described in claim 2, and the step (C) comprises: (C1) the processing unit operating the abnormal time overflow according to an indication of the monitoring unit The logical value of the timeout flag determines whether it is an abnormal operation. If the processing unit determines that the overflow flag is in a second logic, then proceeds to step (E), (C2) if the processing unit Judge that the overflow time flag is in a first logic The processing unit determines that the self operation is abnormal. The method of claim 2, wherein the determining whether the server reset event is caused by a power interruption, the baseboard management controller further comprising an electronic erasing read-only memory electrically connected to the processing unit, wherein the step ( A1) includes: (A12) the processing unit receives the reset request signal, and (A13) the processing unit writes a reset request flag value to the electronically erasable read-only memory, and proceeds to step (A2). The method of determining whether the server reset event is a power interruption as described in claim 5, further comprising the steps (D1) to (D4) before step (B): (D1) the processing unit compares the reset Whether the request flag value meets a set value, if not, step (B) is performed, and (D2) if the result of the step (D1) comparison is YES, the processing unit determines that it is a reset request operation, (D3) The processing unit dumps the reset request flag value to the random access memory, and (D4) the processing unit sends a clear command to clear the reset request flag value of the electronically erasable read-only memory. The method of determining whether the server reset event is a power interruption as described in claim 6, further comprising the step (D5) performed after the step (C): (D5) if the processing unit determines in the step (C) If the operation is not abnormal, the processing unit compares the reset of the random access memory. Whether the request flag value conforms to the set value, and if so, repeats the step (D2); if not, executes step (E). The method of determining whether the server reset event is a power interruption as described in claim 5, further comprising the steps (D1'), (D2'), and (D4') performed after the step (C): (D1) ') The processing unit compares whether the reset request flag value meets a set value, and if not, the process determines that the power reset event, and performs step (E), (D2') if step (D1') If the result of the comparison is yes, then the processing unit determines that it is a reset request operation, and (D4') the processing unit sends a clear command to clear the reset request flag value of the electronically erasable read-only memory. A substrate management controller includes: a random access memory for storing a power reset flag; and a processing unit electrically connected to the random access memory to determine whether to initialize the event according to the received event signal a random access memory; wherein the processing unit further determines, according to a parameter value of the power reset flag, whether it is a contract abnormality, and if the agreement is not abnormal, the processing unit determines whether it is an own operation abnormality, if not If the operation is abnormal, the processing unit determines that it is a power reset event belonging to the server, and the power reset event is related to power interruption to the server. A server that contains: a smart platform management interface for determining whether to output a reset request signal according to a reset command; and a base management controller including a random access memory, storing a power reset flag, a processing unit, and an electrical connection The random access memory determines, according to the received event signal, whether to initialize the random access memory, wherein the event signal includes a reset request signal from the smart platform management interface, and the processing unit further determines a parameter value of the power reset flag to determine whether it is a protocol exception, and an electronic erasure type read only memory, electrically connecting the processing unit, and storing a reset request flag value, the processing unit comparing the weight Whether the request flag value meets a set value, and if the reset request flag value is not equal to the set value, the processing unit determines whether the parameter is abnormal according to the parameter value of the power reset flag, if the agreement is abnormal The processing unit determines whether it is an own operation exception. If the operation is not abnormal, the processing unit determines that the power belongs to the server. Home event, the electric power on reset event is supplied to the server is interrupted.