CN106980468A - Method and device for triggering RAID array reconstruction - Google Patents

Abstract

The application provides a method and a device for triggering RAID array reconstruction, wherein the method is applied to a disk subsystem of a storage device and can comprise the following steps: issuing IO read-write instructions to each physical disk; respectively calculating the average response time of each physical disk based on the response time of the IO read-write instruction returned by each physical disk in a preset statistical period, and respectively judging whether the average response time of the physical disk of each non-fault physical disk reaches the disk abnormal response time threshold corresponding to the model of the physical disk; and marking the physical disk of which the average response time reaches the disk abnormal response time threshold value corresponding to the model as a failed physical disk, and if the failed physical disk belongs to the RAID array, informing the RAID array to which the failed physical disk belongs to reconstruct the failed physical disk. By using the scheme, the accuracy of judging the failed physical disk can be effectively improved.

Description

Method and device for triggering RAID array reconstruction

technical field

The present application relates to the field of computer communication, in particular to a method and device for triggering RAID array reconstruction.

Background technique

RAID array (Redundant Array of Independent Disks, Redundant Array of Independent Disks) is a combination of multiple independent disks (physical disks) in different ways to form a disk group (logical disk), thus providing higher technologies for storage performance and data reliability.

In the field of computer communication, RAID array technology is usually used for redundant protection of data in the disk. When data is written, the data is split into multiple member disks according to the RAID array algorithm. Depending on the level of the RAID array, it can tolerate one or more disk failures or offline. When a disk IO error is detected or the disk is offline, a dedicated hot spare disk or a global hot spare disk can be used to rebuild and restore the data redundancy of the RAID array. .

However, in the existing methods for triggering RAID array reconstruction, only the disk IO error and the disk offline situation are considered, and the slow response time after the disk aging causes business interruption, so how to deal with the slow disk response Triggering the rebuilding of the RAID array becomes an urgent problem to be solved.

Contents of the invention

In view of this, the present application provides a method and device for triggering rebuilding of a RAID array, so as to improve the accuracy of judging a failed physical disk.

Specifically, this application is achieved through the following technical solutions:

According to the first aspect of the present application, a method for triggering RAID array reconstruction is provided, the method is applied to a disk subsystem of a storage device, the storage device includes at least one RAID array, and the RAID array includes several physical disks, The methods include:

Send IO read and write instructions to each physical disk according to the IO read and write requests of each relevant subsystem;

Calculate the average response time of each physical disk based on the response time of the IO read and write commands returned by each physical disk within the preset statistical period;

Determine whether the average response time of the physical disks of each non-faulty physical disk reaches the disk abnormal response time threshold corresponding to its model; wherein, the disk abnormal response time thresholds of different types of physical disks are different;

Mark the physical disk whose average response time reaches the disk abnormal response time threshold corresponding to its model as a faulty physical disk, and if the faulty physical disk belongs to a RAID array, notify the RAID array to which the faulty physical disk belongs to the faulty physical disk to rebuild.

According to the second aspect of the present application, a device for triggering RAID array reconstruction is applied to a disk subsystem of a storage device, the storage device includes at least one RAID array, and the RAID array includes several physical disks, the Said devices include:

The issuing unit is used to issue IO read and write instructions to each physical disk according to the IO read and write requests of each relevant subsystem;

The calculation unit is used to calculate the average response time of each physical disk based on the response time of the IO read and write commands returned by each physical disk within a preset statistical period;

The judging unit is used to respectively judge whether the average response time of the physical disks of each non-faulty physical disk reaches the disk abnormal response time threshold corresponding to its model; wherein, the disk abnormal response time thresholds of different types of physical disks are different;

The marking unit is configured to mark the physical disk whose average response time reaches the disk abnormal response time threshold corresponding to its model as a faulty physical disk, and if the faulty physical disk belongs to a RAID array, then notify the RAID array pair to which the faulty physical disk belongs The failed physical disk is rebuilt.

In the method for triggering RAID array reconstruction proposed in this application, on the one hand, since the disk subsystem can determine the non-faulty physical disk whose average response time reaches the abnormal response time threshold of the disk corresponding to its model based on the average response time of each physical disk Mark the physical disk as a failed physical disk, and notify the RAID array to which the failed physical disk belongs to rebuild, thereby realizing the reconstruction based on the response time of the IO read and write commands of the physical disk to trigger the rebuilding of the RAID array to which the physical disk belongs.

On the other hand, since the response time of each physical disk can be compared with the abnormal response time threshold of the disk corresponding to its model, when judging whether the average response time of a physical disk is abnormal, the IO issued by all services on the disk should be considered comprehensively. For example, the IOs issued by the RAID, the IOs issued by the disk detection task, etc., so as to effectively improve the accuracy of the faulty physical disks marked by the disk subsystem.

Description of drawings

FIG. 1 is a flow chart of a method for triggering RAID array reconstruction shown in an exemplary embodiment of the present application;

FIG. 2 is a hardware structural diagram of a device where a device for triggering RAID array reconstruction is shown in an exemplary embodiment of the present application;

Fig. 3 is a block diagram of an apparatus for triggering rebuilding of a RAID array according to an exemplary embodiment of the present application.

detailed description

Reference will now be made in detail to the exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, the same numerals in different drawings refer to the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with aspects of the present application as recited in the appended claims.

The terminology used in this application is for the purpose of describing particular embodiments only, and is not intended to limit the application. As used in this application and the appended claims, the singular forms "a", "the", and "the" are intended to include the plural forms as well, unless the context clearly dictates otherwise. It should also be understood that the term "and/or" as used herein refers to and includes any and all possible combinations of one or more of the associated listed items.

It should be understood that although the terms first, second, third, etc. may be used in this application to describe various information, the information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of the present application, first information may also be called second information, and similarly, second information may also be called first information. Depending on the context, the word "if" as used herein may be interpreted as "at" or "when" or "in response to a determination."

A RAID array is a technology that combines multiple independent disks (physical disks) in different ways to form a disk group (logical disk), thereby providing higher storage performance and data reliability than a single disk.

In the field of computer communication, RAID array technology is usually used for redundant protection of data in the disk. When data is written, the data is split into multiple member disks according to the RAID array algorithm. Depending on the level of the RAID array, it can tolerate one or more disk failures or offline. When a disk IO error is detected or the disk is offline, a dedicated hot spare disk or a global hot spare disk can be used to rebuild and restore the data redundancy of the RAID array. .

In the related RAID array trigger reconstruction method, when the RAID subsystem receives an IO read and write error returned by a member disk and judges that the error cannot be recovered, it can mark the member disk as faulty and trigger the RAID array to which the member disk belongs reconstruction. In addition, when the RAID subsystem receives a notification message that a member disk is offline, it may also trigger rebuilding of the RAID array to which the offline member disk belongs.

When rebuilding, you can use the hot spare disk to rebuild the faulty disk or offline disk. The RAID subsystem can calculate the data of the corresponding stripe in the hot spare disk according to the RAID array algorithm, and restore the redundancy of the RAID array to which the faulty disk belongs.

Since the disk is a combination of mechanical and electronic devices, affected by factors such as device aging and the environment, in actual applications, disk IO may not return errors but the response time will slow down, which will cause upper-layer applications to read and write the RAID corresponding to the disk. When using an array, the IO return on the disk with slow response time is slower than that of other disks, and the performance of upper-layer applications fluctuates or IO times out. Specifically, when the LUN (logical unit number) created on the RAID array is assigned to the front-end application server for continuous read and write, the performance of the LUN may fluctuate greatly or even the business may be interrupted due to IO timeout. When troubleshooting the large fluctuations in the performance of the LUN, it was found that the status of the RAID array was normal, the status of the disks was also normal, and the member disks did not return IO read and write errors. After further investigation, although the member disks of the RAID array have the same interface and the same speed, the response time of the IO read and write responses returned by some member disks is significantly longer than that of other member disks in the RAID array. After the member disk with long IO read and write response time is removed and the member disk with long IO read and write response time is replaced with a hot spare disk, the performance of the RAID array and LUN returns to normal.

To sum up, the long IO read and write response time of a member disk will seriously affect the performance of the RAID array to which the member disk belongs and the LUN created on the RAID array, resulting in performance fluctuations. In extreme cases, IO timeout may cause service interruption. However, in the existing methods for triggering RAID array reconstruction, only disk IO errors and disk offline situations are considered, and business interruption caused by slow response time after disk aging is not considered.

This application proposes a method for triggering RAID array reconstruction. The disk subsystem of the storage device can issue IO read and write commands to each physical disk according to the IO read and write requests of each related subsystem, and can base on the preset statistics of each physical disk. Calculate the average response time of each physical disk for the response time of the IO read and write commands returned within the period. The disk subsystem can respectively determine whether the average response time of the physical disks of each non-faulty physical disk reaches the abnormal disk response time threshold corresponding to the model of the physical disk. The disk subsystem may mark the physical disk whose average response time reaches the disk abnormal response time threshold corresponding to its model as a faulty physical disk, and if the faulty physical disk belongs to a RAID array, notify the RAID array to which the faulty physical disk belongs rebuild the failed physical disk.

On the one hand, based on the average response time of each physical disk, the disk subsystem can mark the physical disk of the non-faulty physical disk whose average response time reaches the abnormal response time threshold of the corresponding model as a faulty physical disk, and notify the faulty physical disk The RAID array to which the disk belongs is rebuilt, so that the reconstruction of the RAID array to which the physical disk belongs is triggered based on the response time of the IO read and write commands of the physical disk.

On the other hand, since the response time of each physical disk can be compared with the abnormal response time threshold of the disk corresponding to its model, when judging whether the average response time of a physical disk is abnormal, the IO issued by all services on the disk should be considered comprehensively. For example, the IOs issued by the RAID, the IOs issued by the disk detection task, etc., so as to effectively improve the accuracy of the faulty physical disks marked by the disk subsystem.

Referring to FIG. 1, FIG. 1 is a flow chart of a method for triggering RAID array reconstruction shown in an exemplary embodiment of the present application, the method is applied to a disk subsystem of a storage device, and the storage device includes at least one RAID array, The RAID array includes several physical disks, and the method includes:

Step 101: sending IO read and write instructions to each physical disk according to the IO read and write requests of each relevant subsystem;

Step 102: Calculate the average response time of each physical disk based on the response time of the IO read and write commands returned by each physical disk within a preset statistical period;

Step 103: Determine whether the average response time of the physical disks of each non-faulty physical disk reaches the disk abnormal response time threshold corresponding to its model; wherein, the disk abnormal response time thresholds of different types of physical disks are different;

Step 104: mark the physical disk whose average response time reaches the disk abnormal response time threshold corresponding to its model as a faulty physical disk, and if the faulty physical disk belongs to a RAID array, notify the RAID array to which the faulty physical disk belongs to the Rebuild the failed physical disk.

Wherein, the above-mentioned RAID subsystem is used to manage each RAID array in the storage device. For example, the function of the RAID subsystem may include splitting multiple received IO read and write commands based on the RAID array algorithm, and sending the split IO read and write commands to the disk subsystem. The function of the RAID subsystem may also include, when rebuilding the RAID array, based on the algorithm of the RAID array, calculating the data corresponding to the stripe in the hot spare disk, restoring the data redundancy of the RAID array, and the like. Of course, the RAID subsystem also has various functions. Here, the functions of the RAID subsystem are not specifically limited.

The above-mentioned disk subsystem is mainly used to manage all physical disks in the storage device. The disk subsystem is equivalent to the "lower layer" system, which is used to serve the "upper layer" system such as the RAID subsystem. For example, the disk subsystem may scan the physical disks in the storage device, and notify the RAID subsystem of the scanned status information of each physical disk. Of course, in practical applications, the disk subsystem has other functions, and here, the functions of the disk subsystem are not specifically limited.

The aforementioned RAID array refers to a disk group composed of multiple physical disks in the storage device according to the RAID array algorithm. The physical disks in the RAID array can also be called member disks. Depending on the level and implementation of the RAID array, the number of physical disks that can be reconstructed at the same time is also different. For example, the number of physical disks that can be reconstructed at the same time that traditional RAID5 can support is one, and in the implementation of some manufacturers, multiple physical disks can be reconstructed at the same time.

The above response time of the physical disk refers to the time required for the disk subsystem to send an IO read and write command to the physical disk until the physical disk returns the IO read and write response.

The above-mentioned average response time of the physical disk refers to the response time of the completed IO read and write instructions accumulated by the physical disk divided by the number of completed IO read and write instructions accumulated by the physical disk within the preset statistical period is the average response time of the physical disk. Wherein, if the accumulated number of completed IO read and write instructions corresponding to a certain physical disk is zero, the average response time of the physical disk is treated as zero.

The above disk abnormal response time threshold is used to judge whether the response time of the physical disk is abnormal, and when the average response time of the physical disk reaches (greater than or equal to) the abnormal response time threshold, it indicates that the physical disk is abnormal.

It should be noted that the above-mentioned disk abnormality response time threshold may be set by a developer based on actual test data, or may be manually input by a user through an interactive interface. Among them, when the developer sets or the user manually enters the abnormal response time threshold of the disk, if the abnormal response time threshold of the disk is set too large, it may not be possible to accurately detect the physical disk whose response time for IO read and write commands is too long , so that the performance of the RAID array and the LUN created on it cannot be well restored. If the abnormal response time threshold of the disk is set too small, a large number of physical disks with abnormal performance may be detected, and physical disks with normal average response time may be accidentally damaged, resulting in frequent rebuilding of the RAID array and affecting the performance of the storage device. Therefore, in practical applications, developers can set the abnormal response time threshold according to the actual situation. For example, the developer may set the abnormal response time threshold of the disk as the product of the average response time of physical disks of the same model and a preset weighted value of the abnormal response time. Here, the setting of the disk abnormality response time threshold is only exemplified and not specifically limited.

In the embodiment of this application, because the disk subsystem marks the physical disks of the non-faulty physical disks whose average response time reaches the disk abnormal response time threshold corresponding to its model as faulty physical disks based on the average response time of each physical disk, and notifies The RAID to which the failed physical disk belongs is rebuilt, so that the reconstruction of the RAID array to which the physical disk belongs is triggered based on the response time of the IO read and write commands of the physical disk.

In addition, since the response time of each physical disk can be compared with the abnormal response time threshold of the disk corresponding to its model, when judging whether the average response time of a physical disk is abnormal, the IO issued by all services on the disk, such as RAID The IO issued by the disk detection task, the IO issued by the disk detection task, etc., thereby effectively improving the accuracy of the faulty physical disk marked by the disk subsystem.

The above-mentioned method for triggering RAID array reconstruction will be described in detail below.

During implementation, within a preset statistical period, the disk subsystem may issue IO read and write commands to each physical disk according to the IO read and write requests of each relevant subsystem. Each physical disk may return an IO read/write response to the disk subsystem after receiving the IO read/write command and performing corresponding processing.

The disk subsystem can calculate the response time and completed Count the number of IO read and write instructions. At the end of the statistical period, calculate the average response time of each physical disk.

The method for calculating the average response time of a physical disk will be described in detail below by taking the calculation of the average response time of a physical disk as an example.

During implementation, the disk subsystem starts timing when it sends an IO read and write command to a physical disk, and ends the timing after receiving the response of the IO read and write command returned by the physical disk, and can calculate the IO on the physical disk. The response time of the read and write commands, and the response time of this IO read and write command can be accumulated with the response time of the completed IO read and write commands of the physical disk in the current cycle, and the completed The number of IO read and write instructions is increased by 1.

At the end of the current statistics period, the disk subsystem can calculate the average response time of the physical disk in the current statistics period. During implementation, the disk subsystem may divide the accumulated response time of the IO read and write commands of the physical disk by the accumulated number of completed IO read and write commands of the physical disk to obtain the average response time of the physical disk. If the accumulated number of completed IO read and write commands of a certain physical disk in the current statistical period is 0, the average response time of the physical disk is treated as zero.

The calculation method of the average response time of other physical disks is the same as the calculation method described above, and will not be repeated here.

At the end of the current statistical period, the disk subsystem may also calculate the disk abnormal response time threshold corresponding to each type of physical disk.

In an optional implementation manner, the disk subsystem may calculate the product of the average response time of each type of physical disk and a preset abnormal response time weighted value as the disk abnormal response time threshold corresponding to each type of physical disk.

The calculation method of the disk abnormal response time threshold corresponding to a certain type of physical disk is described in detail below.

During implementation, the disk subsystem can accumulate the average response time of the IO read and write commands of multiple physical disks whose average response time is not zero in the current statistical cycle, and can accumulate the average response time Non-zero number of physical disks that are not failed physical disks. Then divide the accumulated average response time of the IO read and write commands of the physical disks of the type of non-faulty physical disks by the accumulated number of non-faulty physical disks whose average response time is not zero to obtain the The average response time of the model's physical disks.

The disk subsystem may multiply the calculated average response time of the type of physical disk by a preset weighted value of abnormal response time to obtain a disk abnormal response time threshold corresponding to the type of physical disk.

Among them, the weighted value of the above-mentioned abnormal response time can be used to judge whether the average response time of the disk is abnormal. It is usually set by the user according to the actual situation and exists in the form of a percentage, such as 200%. Here, the abnormal response time is not weighted Values are specifically defined.

The calculation method of the average response time of the other types of physical disks is the same as the calculation method of the above-mentioned average response time of the physical disk of this type, and will not be repeated here.

In the embodiment of the present application, the disk subsystem may obtain the average response time of each physical disk and the disk abnormal response time threshold of each type of physical disk through the above method. The disk subsystem may determine whether the average response time of the physical disks of each non-faulty physical disk reaches (is greater than or equal to) the abnormal disk response time threshold corresponding to the type of physical disk.

The disk subsystem may mark the physical disk whose average response time reaches the disk abnormal response time threshold corresponding to its model as a faulty physical disk, and if the faulty physical disk belongs to a RAID array, notify the RAID array to which the faulty physical disk belongs Rebuild the failed physical disk.

In order to improve the accuracy of the disk subsystem in detecting a failed physical disk and avoid marking a physical disk with a temporary abnormal response as a failed physical disk, the disk subsystem may not immediately mark the physical disk as a failed physical disk, but record the physical disk in the Judgment results in several consecutive cycles, if the physical disk is determined to be a physical disk whose average response time reaches the abnormal response time threshold of the disk corresponding to its model in several consecutive cycles, the physical disk will be marked as a faulty physical disk If the faulty physical disk belongs to a RAID array, trigger the reconstruction of the RAID array to which the physical disk belongs.

In an optional implementation manner, in order to increase the accuracy and practicability of detecting a faulty physical disk, the above-mentioned several consecutive statistical cycles may be "relatively continuous" several statistical cycles.

When marking, the above-mentioned disk subsystem can respectively record the number of continuous cycles of the physical disk whose average response time reaches the abnormal response time threshold of the disk corresponding to its model; if after several statistical cycles, the average response time reaches the abnormal If the duration number of any disk in the physical disks with the response time threshold reaches the preset duration threshold, the physical disk is marked as a failed physical disk.

When recording, the disk subsystem may, for each physical disk whose average response time reaches the disk abnormal response time threshold corresponding to its model, at the end of the next statistical period, if the physical disk is determined to be the If the physical disk whose average response time reaches the abnormal disk response time threshold corresponding to its model, increase the number of continuous cycles of the physical disk and record it; if the physical disk is not determined as the average response time reaching the abnormal disk corresponding to its model If the response time threshold of the physical disk is exceeded, the number of continuous cycles of the physical disk is reduced and recorded; wherein, the initial value of the number of continuous cycles of the physical disk is zero.

For example, among the above physical disks, if a certain physical disk is determined for the first time as a physical disk whose average response time reaches the disk abnormal response time threshold corresponding to its model, the number of continuous cycles of the physical disk may be set to 1. At the end of the next statistical period, if the physical disk is again determined as the physical disk whose average response time reaches the abnormal response time threshold of the disk corresponding to its model, the duration of the physical disk will be incremented by 1; if the physical disk If the disk is not identified as a physical disk whose average response time reaches the abnormal response time threshold of the disk corresponding to its model, then the number of continuous cycles of the physical disk will be reduced by 1. If the number of continuous cycles of the physical disk is reduced to zero, no Then record the duration of the physical disk.

In another optional implementation manner, the above-mentioned disk subsystem may also mark the physical disk as a failed physical disk based on several "absolutely continuous" statistical periods.

During implementation, the above-mentioned disk subsystem can respectively record the number of continuous cycles of the physical disk whose average response time reaches the abnormal response time threshold of the corresponding model; if after several statistical cycles, the average response time reaches the corresponding If the duration number of any member disk in the physical disk corresponding to the abnormal response time threshold of the model reaches the preset duration threshold, the physical disk is marked as a faulty physical disk.

At the time of recording, for each member disk in a physical disk whose average response time can be determined by the disk subsystem to reach the disk abnormal response time threshold corresponding to its model, at the end of the next statistical period, if the physical disk is determined to be average again If the response time of the physical disk reaches the abnormal response time threshold corresponding to the model, increase the number of continuous cycles of the physical disk and record it; if the physical disk is not found, the average response time reaches the abnormal response time threshold corresponding to the model. If it is a physical disk, the duration of the physical disk is set to zero. Wherein, the initial value of the duration number of the member disk is zero.

In addition, it should be noted that after the RAID array receives the above-mentioned disk subsystem's notification for rebuilding the above-mentioned faulty physical disk, it can determine whether the RAID array currently meets the rebuilding conditions, and if so, rebuild the faulty physical disk; If not, the failed physical disk will not be rebuilt until the RAID array satisfies the rebuilding condition, and then the rebuilding of the failed physical disk will be triggered.

Wherein, the reconstruction conditions may include that the health status of the RAID array meets the reconstruction requirements, there is an available hot spare disk, and the reconstruction of the above-mentioned faulty physical disk will not exceed the number of simultaneously reconstructed physical disks supported by the RAID array to which the faulty physical disk belongs. In practical applications, developers can set the above rebuilding conditions according to the actual situation, and the rebuilding conditions are only described here as examples, and are not specifically limited.

After the disk subsystem completes the relevant judgment and processing of this cycle, the disk subsystem can clear the response time of the completed IO read and write commands and the number of completed IO read and write commands of each physical disk in the current statistical cycle. So that the disk subsystem can collect statistics on these two parameters in the next statistical cycle.

This application proposes a method for triggering RAID array reconstruction. The disk subsystem of the storage device can issue IO read and write commands to each physical disk according to the IO read and write requests of each related subsystem, and can base on the preset statistics of each physical disk. Calculate the average response time of each physical disk for the response time of the IO read and write commands returned within the period. The disk subsystem can respectively determine whether the average response time of the physical disks of each non-faulty physical disk reaches the abnormal disk response time threshold corresponding to the model of the physical disk. The disk subsystem may mark the physical disk whose average response time reaches the disk abnormal response time threshold corresponding to its model as a faulty physical disk, and if the faulty physical disk belongs to a RAID array, notify the RAID array to which the faulty physical disk belongs rebuild the failed physical disk.

On the one hand, based on the average response time of each physical disk, the disk subsystem can mark the physical disk of the non-faulty physical disk whose average response time reaches the abnormal response time threshold of the corresponding model as a faulty physical disk, and notify the faulty physical disk The RAID to which the disk belongs is rebuilt, so that the reconstruction of the RAID array to which the physical disk belongs is triggered based on the response time of the IO read and write commands of the physical disk.

On the other hand, since the response time of each physical disk can be compared with the abnormal response time threshold of the disk corresponding to its model, when judging whether the average response time of a physical disk is abnormal, the IO issued by all services on the disk should be considered comprehensively. For example, the IOs issued by the RAID, the IOs issued by the disk detection task, etc., so as to effectively improve the accuracy of the faulty physical disks marked by the disk subsystem.

In addition, since the disk subsystem can detect the same physical disk as a failed physical disk for several consecutive cycles, it can trigger the rebuilding of the RAID array to which the physical disk belongs, so it can effectively improve the accuracy of the RAID subsystem in detecting failed member disks, and avoid A temporary abnormal response of a member disk occurs.

Corresponding to the foregoing embodiments of the method for triggering RAID array reconstruction, the present application also provides embodiments of an apparatus for triggering RAID array reconstruction.

Embodiments of the apparatus for triggering RAID array rebuilding in the present application may be applied to storage devices. The device embodiments can be implemented by software, or by hardware or a combination of software and hardware. Taking software implementation as an example, as a device in a logical sense, it is formed by reading the corresponding computer program instructions in the non-volatile memory into the memory for operation by the processor of the storage device where it is located. From the hardware level, as shown in Figure 2, it is a hardware structure diagram of the storage device where the device triggering RAID array reconstruction in this application is located, except for the processor, memory, network output interface, and non-volatile memory shown in Figure 2 In addition to the non-volatile memory, the storage device where the device in the embodiment is located usually may also include other hardware according to the actual function of the storage, which will not be repeated here.

Please refer to FIG. 3 , which is a block diagram of an apparatus for triggering rebuilding of a RAID array shown in an exemplary embodiment of the present application. The device is applied to a disk subsystem of a storage device, the storage device includes at least one RAID array, and the RAID array includes several physical disks, and the device includes:

Issuing unit 310, configured to issue IO read and write instructions to each physical disk according to the IO read and write requests of each relevant subsystem;

The calculation unit 320 is configured to calculate the average response time of each physical disk based on the response time of the IO read and write commands returned by each physical disk within a preset statistical period;

The judging unit 330 is used to judge whether the average response time of the physical disks of each non-faulty physical disk reaches the disk abnormal response time threshold corresponding to its model; wherein, the disk abnormal response time thresholds of different types of physical disks are different;

The marking unit 340 is configured to mark the physical disk whose average response time reaches the disk abnormal response time threshold corresponding to its model as a faulty physical disk, and if the faulty physical disk belongs to a RAID array, then notify the RAID array to which the faulty physical disk belongs Rebuild the failed physical disk.

In an optional implementation manner, the abnormal response time threshold of the disk corresponding to the model is the product of the average response time of the physical disk of the model and the preset abnormal response time weighted value;

The judgment unit 330 is specifically used to calculate the number of physical disks of non-faulty physical disks whose average response time of each model is not zero; respectively accumulate the average response time of the physical disks of several non-faulty physical disks of each model; Divide the accumulated average response time of several non-faulty physical disks of each model by the accumulated number of non-faulty physical disks whose average response time is not zero to obtain the average response time of each type of physical disk. Response time: Calculate the product of the average response time of each type of physical disk and the preset abnormal response time weighted value to obtain the disk abnormal response time threshold of each type of physical disk; judge the average response of the physical disk of each non-faulty physical disk Whether the time reaches the disk exception response time threshold corresponding to its model.

In another optional implementation manner, the calculation unit 320 is specifically configured to accumulate the response time of the completed IO read and write commands of each physical disk for the preset statistical period; The number of completed IO read and write commands in the preset statistical period; divide the accumulated response time corresponding to each physical disk by the number of statistical IO read and write commands to obtain the average response time of each physical disk.

In another optional implementation manner, the marking unit 340 is specifically configured to respectively record the number of continuous cycles of the physical disk whose average response time reaches the abnormal response time threshold of the disk corresponding to its model; , the number of continuous cycles of any physical disk among the physical disks whose average response time reaches the disk abnormal response time threshold corresponding to its model reaches the preset continuous cycle threshold, then the physical disk is marked as a faulty physical disk.

In another optional implementation manner, the marking unit 340 is further configured to, for each physical disk whose average response time reaches the disk abnormal response time threshold corresponding to its model, in the next statistical cycle At the end, if the physical disk is determined again as the physical disk whose average response time reaches the abnormal response time threshold of the disk corresponding to its model, then increase the number of continuous cycles of the physical disk and record; if the physical disk is not determined to be If the average response time of the physical disk reaches the abnormal response time threshold of the disk corresponding to its model, the duration number of the physical disk is reduced and recorded; wherein, the initial value of the duration number of the physical disk is zero.

For the implementation process of the functions and effects of each unit in the above device, please refer to the implementation process of the corresponding steps in the above method for details, and will not be repeated here.

As for the device embodiment, since it basically corresponds to the method embodiment, for related parts, please refer to the part description of the method embodiment. The device embodiments described above are only illustrative, and the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in One place, or it can be distributed to multiple network elements. Part or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this application. It can be understood and implemented by those skilled in the art without creative effort.

The above is only a preferred embodiment of the application, and is not intended to limit the application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the application should be included in the application. within the scope of protection.

Claims (10)

1. A method for triggering RAID array reconstruction, characterized in that, the method is applied to a disk subsystem of a storage device, the storage device includes at least one RAID array, and the RAID array includes several physical disks, the method include: Send IO read and write instructions to each physical disk according to the IO read and write requests of each relevant subsystem; Calculate the average response time of each physical disk based on the response time of the IO read and write commands returned by each physical disk within the preset statistical period; Determine whether the average response time of the physical disks of each non-faulty physical disk reaches the disk abnormal response time threshold corresponding to its model; wherein, the disk abnormal response time thresholds of different types of physical disks are different; Mark the physical disk whose average response time reaches the disk abnormal response time threshold corresponding to its model as a faulty physical disk, and if the faulty physical disk belongs to a RAID array, notify the RAID array to which the faulty physical disk belongs to the faulty physical disk to rebuild. 2. The method according to claim 1, wherein the abnormal response time threshold of the disk corresponding to the model is the product of the average response time of the physical disk of the model and the preset abnormal response time weighted value; Said determining whether the average response time of the physical disks of each non-faulty physical disk reaches the abnormal response time threshold of the disk corresponding to its model, including: Calculate the number of physical disks of non-faulty physical disks whose average response time of each model is not zero; Accumulate the average response time of the physical disks of several non-faulty physical disks of each model; Divide the average response time obtained by accumulating the physical disks of several non-failure physical disks of each model by the number of non-failure physical disks whose average response time is not zero corresponding to the model, and obtain the average response of each type of physical disk time; Calculate the product of the average response time of each type of physical disk and the preset abnormal response time weighted value to obtain the disk abnormal response time threshold of each type of physical disk; It is judged whether the average response time of the physical disks of each non-faulty physical disk reaches the abnormal response time threshold of the disk corresponding to its model. 3. The method according to claim 1, characterized in that, calculating the average response time of each physical disk respectively based on the response time of the IO read and write commands returned by each physical disk within a preset statistical period, comprising: accumulating the response time of each physical disk for the completed IO read and write commands of the preset statistical period; Counting the number of completed IO read and write instructions of each physical disk for the preset statistical period; The accumulated response time corresponding to each physical disk is divided by the counted number of IO read and write commands to obtain the average response time of each physical disk. 4. The method according to claim 1, wherein marking the physical disk whose average response time reaches the abnormal disk response time threshold corresponding to its model as a failed physical disk comprises: Record the number of continuous cycles of the physical disk whose average response time reaches the abnormal response time threshold of the disk corresponding to its model; If after several statistical periods, the number of continuous periods of any physical disk whose average response time reaches the abnormal response time threshold of the disk corresponding to its model reaches the preset continuous period threshold, mark the physical disk as is a failed physical disk. 5. The method according to claim 4, wherein the recording of the number of continuous cycles of the physical disk whose average response time reaches the abnormal disk response time threshold corresponding to its model respectively includes: For each physical disk whose average response time reaches the disk abnormal response time threshold corresponding to its model, at the end of the next statistical period, if the physical disk is determined again as the average response time reaching the threshold corresponding to its model If the physical disk whose average response time reaches the abnormal response time threshold of the corresponding model of the physical disk is not identified as the physical disk whose average response time reaches the abnormal response time threshold of the corresponding model, then Decrease the duration number of the physical disk and record it; wherein, the initial value of the duration number of the physical disk is zero. 6. A device for triggering RAID array reconstruction, characterized in that the device is applied to a disk subsystem of a storage device, the storage device includes at least one RAID array, and the RAID array includes several physical disks, the device include: The issuing unit is used to issue IO read and write instructions to each physical disk according to the IO read and write requests of each relevant subsystem; The calculation unit is used to calculate the average response time of each physical disk based on the response time of the IO read and write commands returned by each physical disk within a preset statistical period; The judging unit is used to respectively judge whether the average response time of the physical disks of each non-faulty physical disk reaches the disk abnormal response time threshold corresponding to its model; wherein, the disk abnormal response time thresholds of different types of physical disks are different; The marking unit is configured to mark the physical disk whose average response time reaches the disk abnormal response time threshold corresponding to its model as a faulty physical disk, and if the faulty physical disk belongs to a RAID array, then notify the RAID array pair to which the faulty physical disk belongs The failed physical disk is rebuilt. 7. The device according to claim 6, wherein the abnormal response time threshold of the disk corresponding to the model is the product of the average response time of the physical disk of the model and the preset abnormal response time weighted value; The judgment unit is specifically used to calculate the number of physical disks of non-failure physical disks whose average response time of each model is not zero; respectively accumulate the average response time of the physical disks of several non-failure physical disks of each model; The average response time of physical disks accumulated by several non-failure physical disks of each model is divided by the number of non-failure physical disks whose average response time is not zero corresponding to the model, to obtain the average response time of each type of physical disk; Calculate the product of the average response time of each type of physical disk and the preset abnormal response time weighted value to obtain the disk abnormal response time threshold of each type of physical disk; determine whether the average response time of each non-faulty physical disk has reached The disk exception response time threshold corresponding to its model. 8. The device according to claim 6, wherein the calculation unit is specifically configured to accumulate the response time of each physical disk for the completed IO read and write instructions of the preset statistical cycle; count each physical disk The number of completed IO read and write instructions for the preset statistical period; the cumulative response time corresponding to each physical disk and the number of statistical IO read and write instructions are divided to obtain the average Response time. 9. The device according to claim 6, wherein the marking unit is specifically used to respectively record the number of continuous cycles of the physical disk whose average response time reaches the abnormal response time threshold of the disk corresponding to its model; After the statistical period, if the average response time of the physical disks whose average response time reaches the abnormal response time threshold of the disk corresponding to its model and the number of continuous periods of any physical disk reaches the preset continuous period threshold, the physical disk is marked as a failed physical disk . 10. The device according to claim 9, wherein the marking unit is further configured to, for each physical disk whose average response time reaches the disk abnormal response time threshold corresponding to its model, set the following At the end of a statistical period, if the physical disk is again determined as the physical disk whose average response time reaches the abnormal response time threshold of the disk corresponding to its model, then increase the number of continuous cycles of the physical disk and record it; if the physical disk does not If it is determined that the average response time of the physical disk reaches the abnormal response time threshold of the disk corresponding to its model, the number of continuous cycles of the physical disk is reduced and recorded; wherein, the initial value of the number of continuous cycles of the physical disk is zero.