WO2012119432A1 - Method for improving stability of computer system, and computer system - Google Patents

Abstract

The present invention relates to a method for improving the stability of a computer system and a computer system. The method for improving the stability of a computer system includes: when a computer system is started or is running, collecting the error data generated by the devices in the computer system; storing the error data in a non-volatile memory; when the computer system is restarted, performing state recovery on the devices generating the error data according to the error data. Recording the error data into a non-volatile memory when the computer system is running, and reading the error data in the non-volatile memory to perform state recovery on corresponding devices in the computer system during reboot, solves the problem of degraded system stability, which is directly caused by failure of processing such as disenabling and isolation carried out by the computer system on some abnormal devices or devices predicted to be damaged because of device state initialization after the computer system is restarted, improving the stability of the computer system.

Description

 Method and computer system for improving stability of computer system

 The present invention relates to computer technology, and more particularly to a method and computer system for improving the stability of a computer system. Background technique

 Because the high-end fault-tolerant computer system carries Mission Critical in the financial, telecommunications, aerospace, power and other industries, it needs to ensure 24 hours a day, 365 days of continuous operation, and ensure the correctness of the data, so it needs to have a high degree of stability. Availability Availability and Serviceability (RAS) features. Specifically, stability requires the computer to continue to operate, automatically detecting and correcting system errors. Availability requires that critical resources of the computer system be backed up, able to detect potential problems, and be able to transfer tasks that are running on them to backup resources to keep the computer system up and running down and downtime. Serviceability requirements The computer system is capable of real-time online diagnosis, pinpointing the underlying problem, and making accurate and accurate repairs.

 In the prior art, device error data of a running computer system is usually collected through an Onboard Administrator (OA) to utilize these erroneous data for fault prediction. When the number of device failures reaches the set threshold, enable the backup device or perform a hot replacement. These error data severely affect the stability of the restarted computer system or the device that has been offline and then turned on again. Summary of the invention

The embodiment of the invention provides a method and a computer system for improving the stability of a computer system to improve the stability of the computer system. An embodiment of the present invention provides a method for improving stability of a computer system, including: collecting, when the computer system is started or running, erroneous data generated by a device of the computer system;

 Storing the erroneous data in a non-volatile memory;

 When the computer system is restarted, state recovery processing is performed on the device that generates the erroneous data based on the error data.

 The embodiment of the present invention further provides a method for improving the stability of a computer system, including: collecting, during operation, an abnormality information of a device in which an abnormality occurs in a device of the computer system;

 Storing the abnormality information of the device in which the abnormality occurs in the non-volatile memory;

 The computer system performs state restoration on the off-line device requesting to go online again according to the abnormality information.

 The embodiment of the invention further provides a computer system, including:

 An error collection unit, configured to collect error data generated by a device of the computer system when the computer system is running or started;

 a storage unit, configured to store the error data into a non-volatile memory;

 And a recovery processing unit, configured to perform state recovery processing on the device that generates the erroneous data according to the error data when the computer system is restarted.

 The embodiment of the invention further provides a computer system, including:

 An abnormality information collecting unit, configured to collect abnormal information of a device in which an abnormality occurs in a device of the computer system when the computer system is running;

 a storage unit, configured to store the abnormality information of the device in which the abnormality occurs in the non-volatile memory;

 The state recovery unit is configured to perform state recovery on the offline device that is requested to be re-online according to the abnormality information.

A method and a computer system for improving stability of a computer system provided by an embodiment of the present invention The error data is recorded into the non-volatile memory during the operation of the computer system, and the error data in the non-volatile memory is read during the restart process to perform state recovery processing on the corresponding device in the computer system, thereby solving the computer system. The failure of the computer system before the restart of the device state to disable or isolate the device that is abnormal or predicted to be damaged may directly lead to the problem of lowering the stability of the system and improve the stability of the computer system. DRAWINGS

 1 is a flow chart of a method for improving stability of a computer system according to an embodiment of the present invention;

 2 is a flow chart of another method for improving stability of a computer system according to an embodiment of the present invention;

 3 is a schematic diagram of device state recovery in a case where a computer system is restarted in a method for improving stability of a computer system according to an embodiment of the present invention;

 4 is a schematic diagram of a BIOS policy configuration menu of a computer system in a method for improving stability of a computer system according to an embodiment of the present invention;

 5 is a flowchart of recovering a DIMM isolation state in a method for improving stability of a computer system according to an embodiment of the present invention;

 FIG. 6 is a flowchart of a processor core disable recovery process in a method for improving stability of a computer system according to an embodiment of the present invention;

 FIG. 7 is a flowchart of a state recovery process of cache disable information in a method for improving stability of a computer system according to an embodiment of the present invention; FIG.

 FIG. 8 is a flowchart of a state recovery process of a node that has gone offline abnormally in a method for improving stability of a computer system according to an embodiment of the present invention; FIG.

 FIG. 9 is a schematic structural diagram of a computer system according to an embodiment of the present invention;

FIG. 10 is a schematic structural diagram of another computer system according to an embodiment of the present invention. detailed description

 The present invention will be further described in detail below with reference to the accompanying drawings.

 FIG. 1 is a flowchart of a method for improving stability of a computer system according to an embodiment of the present invention. This embodiment performs state recovery on the devices in the computer system for the erroneous data collected by the restarted computer system during the last run to improve the stability of the computer system. As shown in Figure 1, the method includes:

 Step 11: The computer system collects error data generated by the device of the computer system when the computer system is started or running; the error data may be abnormal information of the device, isolation information of dual inline memory modules (DIMMs), Disable information for the processor core, cache disable information, and so on.

 Step 12. Store the information in a Non-Volatile Memory (NVM).

 When the error data is abnormal information of the device, the computer system stores the abnormality information of the device in which the abnormality occurs in the non-volatile memory at runtime;

 The method further includes: the computer system performing state restoration on the off-line device requesting the re-online according to the abnormality information during operation.

 When the error data is DIMM isolation information, the computer system determines whether the DIMM is replaced at runtime, and if so, clears the DIMM isolation information stored in the non-volatile memory; otherwise, The DIMM is isolated when the computer system is restarted.

 Step 13: When the computer system restarts, perform state recovery processing on the device that generates the error data according to the error data.

For example, the computer system isolates corresponding DIMMs in the computer system based on DIMM isolation information. For example, the computer system prohibits the corresponding processor core in the computer system from participating in the selection of the Processor Boot Strap Processor (PBSP) according to the disable information of the processor core, or disables the corresponding processor in the computer system. nuclear. Computer system Re-disabling the corresponding cache in the computer system based on the cache disable information.

 In this embodiment, the computer system saves the error data collected by the runtime in the NVM, and performs state recovery processing on the corresponding device according to the error data during the restart, thereby avoiding the problem that the computer system will be problematic or unstable after the initialization. The device is enabled as a normal device, improving the stability of the computer system.

 FIG. 2 is a flow chart of another method for improving stability of a computer system according to an embodiment of the present invention. This embodiment separately deals with the problem that the abnormality of the computer system in the running computer system causes the computer system to be unstable after being offline. As shown in Figure 2, the method includes:

 Step 21: When the computer system is running, collecting abnormal information of the device that is abnormal in the device of the computer system;

 Step 22: The abnormality information of the device in which the abnormality occurs is stored in the non-volatile memory. Step 23: The computer system performs state restoration on the offline device that requests the re-online according to the abnormality information.

 After the computer system stores the abnormality information of the device in which the abnormality occurs in the non-volatile memory, the computer system further includes: the computer system determines whether the device is replaced, and if yes, deletes the nonvolatile The exception information in the memory; otherwise, the state recovery is performed.

 In this embodiment, the computer system restores the status of the corresponding offline device that has been re-requested online according to the abnormal information stored in the NVM, and avoids the system that the device that is offline due to the abnormality of the device is re-lived as a normal device. Unstable, improving the stability of the computer system.

 For example, when the computer system is running, if the device fails or is disabled by the system, its Basic Input Output System (BIOS) will save this information in the NVM. When the computer system is restarted, the information is analyzed and processed. Before the unstable device is started, the system is configured according to the information, the unstable device is isolated or disabled, and the unstable device in the computer system is restored to the state before the restart. Ensure the stability of the computer system.

FIG. 3 is a schematic diagram of device state recovery in a case where a computer system is restarted in a method for improving stability of a computer system according to an embodiment of the present invention. As shown in Figure 3, when the computer system is running, it will receive The error data such as the DIMM isolation information, the processor core health status information, and the cache disable information are stored in the NVM. When the computer system is restarted, the BIOS configuration policy is enabled, that is, the device state recovery function is enabled in the BIOS configuration interface, and the error is obtained from the NVM. Data, according to the corresponding error data, the corresponding device state recovery distribution program is called to perform state recovery processing on the corresponding device.

 Further, the BIOS configuration interface can also be provided with switches to enable the user to perform flexible device state recovery according to requirements. As shown in Figure 4, the configuration policy switches of different devices are enabled as required. Only the devices in the enabled state will be restored according to the corresponding policies. When the reboot starts and the system is initialized, the system enters the device status recovery distribution.

 The DIMM isolation information, the processor core health status information, and the cache disabled information shown in FIG. 4 are taken as an example to further explain the method for improving the stability of the computer system.

 When these error data are generated, the BIOS saves the error data in non-volatile memory. When the computer system is restarted for initialization, the device state recovery distribution program is called to obtain the error data from the NVM, according to different strategies. deal with.

 Specifically, the DIMM isolation state recovery process is as shown in FIG. 5, including:

 Step 51. When the computer system is running, the error check and correction of a certain DIMM (Error

Correcting Code, ECC) The error reaches the specified threshold;

 Step 52: The computer system identifies the source of the error through the Log information;

 Step 53: The computer system marks the DIMM as an impending failure state, and disables the DIMM;

 Step 54: The computer system saves the invalid DIMM information to the NVM.

 Step 55, the computer system determines whether the DIMM is replaced, and if so, step 58 is performed; otherwise, step 56;

 Step 56: After the computer system is restarted, the DIMM information stored in the NVM is read. Step 57: The computer system calls the device state recovery distribution program to isolate the corresponding DIMM, and the computer system continues to boot, and the operating system (OS) is booted.

Step 58, the computer system clears the invalid DIMM information in the NVM, and the computer system recovers For a healthy state.

 In this embodiment, the DIMM in the computer system protects the data by using a mechanism such as ECC. When the number of ECC errors is found to reach the set upper threshold, the system marks the DIMM as a failed state and disables the failure information. Save in NVM. During the period from the normal operation of the computer system to the restart, if the DIMM is replaced by the hot plug process, the failure information of the DIMM in the NVM is cleared. Otherwise, when the system restarts, the device status recovery dispatcher is called during startup, reading the failure information of all DIMMs in the system, and isolating it using the configuration program.

 The processor core disable recovery process is shown in Figure 6, including:

 Step 61: The computer system detects the health status of the processor core when the computer is started for the first time; Step 62: The computer system disables the processor core that fails the detection;

 Step 63: The computer system records the disable information to the NVM;

 Step 64, PBSP selection, the computer system continues to boot, booting the OS;

 Step 65: When the computer system is started for the second time, the device state recovery distribution program is invoked, and the disable information is read from the NVM;

 Step 66: The computer system prohibits the corresponding processor core from participating in the PBSP selection according to the read disable information, or directly disables the corresponding processor core;

 Step 67: The computer system continues to start, detecting the health status of the processor core;

 Step 68: The computer system disables the undetected processor core;

 Step 69: The computer system records the disabling information into the NVM.

After that, the computer system performs PBSP selection, and the computer system continues to boot, booting the OS. Some unstable cores may be detected when the system starts up. The core will be blocked during this startup, but the next time it starts, the core may still pass the detection, further competing PBSP and SBSP to become the main thread of the system. At this time, if there is a problem with the core, the entire system will crash immediately. In this process, the information of the unstable core detected is stored in the NVM, and the device state recovery process is used to identify and limit the unstable core to participate in the PBSP competition or directly disable the core. The state recovery processing flow for the cache disable information is as shown in FIG. 7, and includes: Step 71: In the computer system, when the number of ECC errors of a cache (Cacheline) reaches a threshold, the computer system disables the cache.

 Step 72: The computer system records the disable information to the NVM;

 Step 73: When the computer system restarts, the device state recovery distribution program is invoked, and the cache disable information is read;

 Step 74: The computer system uses the system service to reconfigure the cache.

 After that, the computer system continues to boot, booting the OS.

 The cache security technology stipulates that when a Cacheline ECC error reaches the threshold, it will be disabled by the system. This process saves this information through the NVM. When the next computer system starts, the device status recovery distribution program is called to read this information. Cacheline configuration, re-close the Cacheline that was once marked as disabled.

 In the running computer system, the information about the abnormality of the device is saved to the NVM. Suppose, in a scenario, the user makes a offline request for an abnormal node in the running computer system due to energy saving or other requirements. After the computer system completes the offline operation, the user requests the node that has been offline to go online again due to the service requirement. At this time, the system will call the device state recovery program to recover the node status. As shown in Figure 8, the following includes:

 Step 81: When an abnormality occurs in the device in the running computer system, the abnormal condition is processed. Step 82: Keep the abnormal information in the NVM.

 Step 83: The user needs to drop a node or device offline;

 Step 84, the offline operation is completed, and the computer system continues to run;

 Step 85: The user needs to bring the offline node or device back online;

 Step 86: The computer system invokes a device state recovery distribution program in the online operation, and performs a state recovery operation on each device;

 Step 87, the device is online;

Step 88, the computer system continues to run. This recovery operation is optional. As shown in Figure 4, the switch set in the BIOS configuration interface can be configured for online operation.

 The biggest difference between this processing flow and the above processing flow is that the computer system restart is not involved, and the whole operation is dynamic.

 The foregoing method embodiment solves the problem that the computer system is restarted by recording the error data into the NVM during the running of the computer system, and reading the error data in the NVM during the restarting process to perform state recovery processing on the corresponding device in the computer system. The failure of the computer system to disable or isolate the abnormality or predicted damage of the device caused by the initialization of the device state directly leads to the problem of the stability of the system, which improves the stability of the computer system. And, the abnormal information is saved in the NVM in the online process of running the computer system, and the device is restored according to the abnormal information saved by the NVM during the re-up of the device, thereby solving the running process of the computer system. The user requests the device to go offline after being offline. At this time, the state before the device goes offline is initialized. The unstable part (possibly another child device) that is disabled in the device may run in the system again, resulting in system stability. The problem of decline has improved the stability of the computer system.

 A person skilled in the art can understand that all or part of the steps of implementing the above method embodiments may be completed by program instructions and related hardware, and the foregoing program may be stored in a computer readable storage medium, when executed, The foregoing steps include the steps of the foregoing method embodiments; and the foregoing storage medium includes: a medium that can store program codes, such as a ROM, a RAM, a magnetic disk, or an optical disk.

 FIG. 9 is a schematic structural diagram of a computer system according to an embodiment of the present invention. As shown in FIG. 9, the computer system includes: an error collecting unit 91, a storage unit 92, and a recovery processing unit 93.

The error collection unit 91 is configured to collect error data generated by the device of the computer system when the computer system is running or starting up; the storage unit 92 is configured to store the error data into the NVM; and the recovery processing unit 93 is configured to When the computer system is restarted, the root performs state recovery processing on the device that generates the erroneous data according to the error data. When the error data is abnormal information of the device, the recovery processing unit 93 is further configured to perform state recovery on the offline device that is requested to be re-online according to the abnormal information when the computer system is running. .

 The recovery processing unit 93 is specifically configured to isolate the corresponding DIMMs in the computer system according to the DIMM isolation information.

 The computer system may further include: a replacement determination unit and an information removal unit.

 The replacement determining unit is configured to determine whether the DIMM is replaced when the computer system is running; the information clearing unit is configured to: when the replacement determining unit determines that the DIMM is replaced, to be stored in the non-volatile memory The DIMM isolation information is cleared. Correspondingly, the recovery processing unit is configured to isolate the DIMM when the computer system is restarted if the replacement determination unit determines that the DIMM is not replaced.

 The recovery processing unit 93 may be specifically configured to prohibit the corresponding processor core in the computer system from participating in the selection of the PBSP according to the disable information of the processor core, or to disable the corresponding processor core in the computer system.

 As another example, the recovery processing unit 93 may be specifically configured to re-close the corresponding cache in the computer system according to the cache disable information. The error data collected by the error collection unit and the recovery processing operation performed by the recovery processing unit are specifically described in the above method embodiments.

 In this embodiment, the computer system stores the error data collected by the runtime in the NVM through the error collection unit, and performs state recovery processing on the corresponding device according to the error data by the recovery processing unit during the restart, thereby avoiding the computer system having the initialization after the initialization. A problematic or unstable device is enabled as a normal device, improving the stability of the computer system.

 FIG. 10 is a schematic structural diagram of another computer system according to an embodiment of the present invention. As shown in FIG. 10, the computer system includes: an abnormality information collecting unit 101, a storage unit 102, and a state restoring unit 103.

The abnormality information collecting unit 101 is configured to collect abnormality information of a device in which an abnormality occurs in a device of the computer system when the computer system is running; the storage unit 102 is configured to: The abnormality information is stored in the NVM; the state restoring unit 103 is configured to perform state recovery on the offline device that is requested to be re-online according to the abnormality information.

 The computer system provided by the embodiment of the present invention may further include: a replacement judging unit and an information deleting unit.

 The replacement determining unit is configured to determine whether the device is replaced after the abnormality information of the device in which the abnormality occurs is stored in the non-volatile memory when the computer system is running; the information deleting unit is configured to determine the replacement Determining, by the unit, that the device is replaced, deleting the abnormality information in the non-volatile memory; and correspondingly, the state recovery unit is configured to: if the replacement determining unit determines that the device is not replaced, Perform the state recovery.

 In this embodiment, the state recovery unit performs a state recovery on the corresponding offline device that is re-requested by the state recovery unit according to the abnormality information stored in the NVM, thereby avoiding that the device that is offline due to the abnormality of the device is re-online as the normal device. The resulting system is unstable and improves the stability of the computer system.

 Each unit in the above system embodiment can be set in the BIOS.

 It should be noted that the above embodiments are only for explaining the technical solutions of the present invention, and are not intended to be limiting; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those skilled in the art that: The technical solutions described in the foregoing embodiments are modified, or some of the technical features are equivalently replaced. The modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

Claim  A method for improving the stability of a computer system, comprising:  The computer system collects error data generated by the device of the computer system when it is started or running;  Storing the erroneous data in a non-volatile memory;  When the computer system is restarted, state recovery processing is performed on the device that generates the erroneous data based on the error data.  2. The method for improving the stability of a computer system according to claim 1, wherein the error data is abnormal information of the device;  The method further includes: performing, during operation, the state recovery of the off-line device requesting to go online according to the abnormality information.  The method for improving the stability of a computer system according to claim 1 or 2, wherein the error data is DIMM isolation information of the dual in-line storage module;  The process of performing state recovery processing on the corresponding device in the computer system according to the error data includes:  Corresponding DIMMs in the computer system are isolated based on the DIMM isolation information. 4. The method of improving stability of a computer system according to claim 3, wherein said computer system determines whether said DIMM is replaced during operation, and if so, is stored in said non-volatile memory The DIMM isolation information is cleared; otherwise, the DIMM is isolated when the computer system is restarted.  The method for improving the stability of a computer system according to claim 1 or 2, wherein the error data is a disable information of a processor core;  The process of performing state recovery processing on the corresponding device in the computer system according to the error data includes: The corresponding processor cores in the computer system are prohibited from participating in the selection of the PBSP according to the disable information of the processor core, or the corresponding processor cores in the computer system are disabled. 6. The method of improving stability of a computer system according to claim 1 or 2, wherein the error data is cache disable information;  The process of performing state recovery processing on the corresponding device in the computer system according to the error data includes:  Re-closing the corresponding cache in the computer system according to the cache disable information.  7. A method for improving the stability of a computer system, comprising:  While the computer system is running, collecting abnormal information of the device in the device of the computer system that is abnormal;  Storing the abnormality information of the device in which the abnormality occurs in the non-volatile memory;  The computer system performs state restoration on the off-line device requesting to go online again according to the abnormality information.  The method for improving the stability of a computer system according to claim 7, wherein the computer system, after the operating system stores the abnormality information of the abnormal device in the non-volatile memory, further includes: The computer system determines whether the device is replaced, and if so, deletes the abnormal information in the non-volatile memory; otherwise, performs the state recovery.  9. A computer system, comprising:  An error collection unit, configured to collect error data generated by a device of the computer system when the computer system is running or started;  a storage unit, configured to store the error data into a non-volatile memory;  And a recovery processing unit, configured to perform state recovery processing on the device that generates the erroneous data according to the error data when the computer system is restarted.  The computer system according to claim 9, wherein the error data is abnormal information of the device;  The recovery processing unit is further configured to perform state recovery on the off-line device that requests to be re-online according to the abnormality information when the computer system is running. The computer system according to claim 9 or 10, wherein the number of errors According to the DIMM isolation information of the dual in-line storage module, the recovery processing unit is specifically configured to isolate the corresponding DIMMs in the computer system according to the DIMM isolation information.  The computer system according to claim 11, further comprising: a replacement determining unit, configured to determine whether the DIMM is replaced when the computer system is running;  An information clearing unit, configured to: if the replacement determining unit determines that the DIMM is replaced, clear the DIMM isolation information stored in the non-volatile memory;  The recovery processing unit is configured to isolate the DIMM when the computer system is restarted, if the replacement determining unit determines that the DIMM is not replaced.  The computer system according to claim 9 or 10, wherein the error data is a disable information of a processor core; the recovery processing unit is specifically configured to disable the computer system according to a disable information of a processor core. The corresponding processor core participates in the selection of the PBSP or disables the corresponding processor core in the computer system.  The computer system according to claim 9 or 10, wherein the error data is cache disable information; the recovery processing unit is specifically configured to re-close the corresponding cache in the computer system according to the cache disable information.  15. A computer system, comprising:  An abnormality information collecting unit, configured to collect abnormal information of a device in which an abnormality occurs in a device of the computer system when the computer system is running;  a storage unit, configured to store the abnormality information of the device in which the abnormality occurs in the non-volatile memory;  The state recovery unit is configured to perform state recovery on the offline device that is requested to be re-online according to the abnormality information. The computer system according to claim 15, further comprising: a replacement determining unit, configured to store, in the non-volatile memory, abnormal information of a device in which an abnormality occurs when the computer system is running After that, it is determined whether the device is replaced; An information deleting unit, configured to delete the abnormality information in the non-volatile memory if the replacement determining unit determines that the device is replaced;  The state restoration unit is configured to perform the state recovery if the replacement determination unit determines that the device is not replaced.