CN107239238A - A kind of I/O operation method and device of the storage based on distributed lock - Google Patents

Abstract

The invention discloses a storage IO operation method based on a distributed lock, which is applied to a first host. The first host and multiple other hosts are connected to the same shared storage, and each other is read according to a preset reading cycle. The host writes the time stamp to the shared storage according to the preset write cycle. When issuing the first IO operation command to the target data in the shared storage, it detects whether the target data is occupied by the second host, and if so, reads The timestamp of the second host, according to the timestamp of the second host read this time and the timestamp of the second host read last time, determine whether the second host is offline, if yes, release the second host from the target data Occupied and perform IO operations. Applying the technical solution provided by the present invention avoids the situation that the second host is not offline, but is accidentally isolated due to IO timeout of the second host. The invention also discloses an IO operation device based on distributed lock storage, which has corresponding technical effects.

Description

A storage IO operation method and device based on distributed locks

technical field

The present invention relates to the field of computer application technology, in particular to a storage IO operation method and device based on distributed locks.

Background technique

With the development of computer technology and network technology, virtualized large-scale clusters often appear in the actual production environment of storage. In a virtualization system, there may be multiple hosts of different systems or the same system connected to the same resource or a group of resources. At this time, distributed locks are needed to avoid mutual interference among hosts acquiring resources.

In the prior art, a shared storage solution is usually used to implement distributed locks. For each host connected to the same shared storage, when one of the hosts occupies the shared storage, for example, when one of the hosts writes to the shared storage, it is called host No. The IO operation on the shared storage can only be performed after the writing operation process is completed. Usually, a timeout mechanism is adopted in the distributed lock scheme of the prior art, that is, when host No. 1 is disconnected from the shared storage for a certain period of time, the distributed lock will release host No. Hosts can occupy this shared storage. However, if the IO pressure of the shared storage is high, even if host No. 1 has not disconnected from the shared storage, the distributed lock will mistakenly release host No. 1 from occupying the shared storage. When the IO operation command is issued, because other hosts occupy the shared storage at this time, host No. 1 is isolated and cannot complete the previously unfinished IO operations.

Therefore, how to find a stable distributed lock solution is a technical problem urgently needed by those skilled in the art.

Contents of the invention

The purpose of the present invention is to provide an IO operation method and device based on distributed lock storage, which avoids that the second host is not offline, but due to the unexpected isolation of the second host due to the IO timeout of the second host, it cannot The case where the IO operation to the shared storage is completed.

In order to solve the above technical problems, the present invention provides the following technical solutions:

An IO operation method based on distributed lock storage, applied to a first host, the first host and multiple other hosts are connected to the same shared storage, the method comprising:

reading the time stamp written by each other host to the shared storage according to the preset write cycle according to the preset read cycle;

When issuing a first IO operation instruction to the target data in the shared storage, detecting whether the target data is occupied by a second host, where the second host is any one of a plurality of other hosts;

If so, read the timestamp of the second host, and determine the second host according to the timestamp of the second host read this time and the timestamp of the second host read last time whether offline;

If so, release the occupation of the target data by the second host and perform an IO operation.

Preferably, the determining whether the second host is offline according to the timestamp of the second host read this time and the timestamp of the second host read last time includes:

Obtain the IO waiting time of the second host;

Determine whether the timestamp of the second host read last time plus the IO waiting time is greater than the timestamp of the second host read this time;

If not, it is determined that the second host is offline.

Preferably, the obtaining the IO waiting time of the second host includes:

determining the IO waiting time of the second host according to the waiting time of the second IO command queue of the second host;

Wherein, the second IO instruction queue has a non-timeout attribute, and includes a second IO operation instruction issued by the second host to the target data.

Preferably, the first IO operation instruction is an instruction in the first IO instruction queue with a non-timeout property.

Preferably, the read period is greater than or equal to the write period.

An IO operation device based on distributed lock storage, applied to a first host, the first host and multiple other hosts are connected to the same shared storage, the device includes:

A time stamp reading module, configured to read the time stamp written by each other host to the shared storage according to the preset write cycle according to the preset read cycle;

The target data detection module is used to detect whether the target data is occupied by the second host when issuing the first IO operation instruction to the target data in the shared storage, and if so, trigger the second host offline determination module, The second host is any one of multiple other hosts;

The second host offline determination module is configured to read the timestamp of the second host, according to the timestamp of the second host read this time and the timestamp of the second host read last time , determining whether the second host is offline, and if so, triggering the IO operation module;

The IO operation module is configured to release the occupation of the target data by the second host and perform an IO operation.

Preferably, the module for determining that the second host is offline includes:

The IO waiting time obtaining submodule is used to obtain the IO waiting time of the second host;

The timestamp determination submodule is used to determine whether the timestamp of the second host read last time plus the IO waiting time is greater than the timestamp of the second host read this time, if not, enter The second host determines the submodule offline;

The second host offline determining submodule is configured to determine that the second host is offline.

Preferably, the IO waiting time acquisition submodule is specifically used for:

determining the IO waiting time of the second host according to the waiting time of the second IO command queue of the second host;

Wherein, the second IO instruction queue has a non-timeout attribute, and includes a second IO operation instruction issued by the second host to the target data.

Preferably, the first IO operation instruction is an instruction in the first IO instruction queue with a non-timeout property.

Preferably, the read period is greater than or equal to the write period.

By applying the technical solution provided by the embodiment of the present invention, the time stamp written by each other host to the shared storage according to the preset write cycle is read according to the preset read cycle, and the target data in the shared storage is issued During the first IO operation instruction, detect whether the target data is occupied by the second host, if yes, read the timestamp of the second host, according to the timestamp of the second host read this time and the second read last time The time stamp of the host, to determine whether the second host is offline, and if so, release the occupation of the target data by the second host and perform IO operations.

When the reading and writing pressure of the shared storage is relatively high, it may not be able to respond to the IO operation of the second host in time. Therefore, if the IO operation command issued by the second host has an IO timeout, it may be that the second host has disconnected from the shared storage. The connection causes the IO timeout of the second host, or the IO pressure of the shared storage may cause the IO timeout of the second host. In the solution of the present invention, when the first host detects that the target data is occupied by the second host, it reads the time stamp of the second host, and compares the time stamp of the second host read this time with the time stamp of the second host read last time. The host's timestamp is compared to determine whether the second host is offline. When the IO pressure of the shared storage is high, the IO of the second host will time out, but the second host can send a time stamp to make the first host determine whether the second host is offline, avoiding that the second host is not offline, but due to The IO timeout of the second host causes unexpected isolation of the second host, so that the IO operation on the shared storage cannot be completed.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments of the present invention. Those skilled in the art can also obtain other drawings based on these drawings without creative work.

Fig. 1 is a schematic diagram of an implementation scenario of an IO operation method based on distributed lock storage in the present invention;

Fig. 2 is the implementation flowchart of a kind of IO operation method based on the storage of distributed lock in the present invention;

FIG. 3 is a schematic structural diagram of an IO operation device based on distributed lock storage in the present invention.

detailed description

The core of the present invention is to provide an IO operation method based on distributed lock storage. The first host determines whether the second host is offline by reading the timestamp of the second host, which avoids that the second host is not offline, but due to the second The IO timeout of the second host causes unexpected isolation of the second host, so that the IO operation on the shared storage cannot be completed.

In order to enable those skilled in the art to better understand the solution of the present invention, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. Apparently, the described embodiments are only some of the embodiments of the present invention, but not all of them. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

Please refer to FIG. 1 , which is a schematic diagram of an implementation scenario of an IO operation method based on distributed lock storage in the present invention. The host 101 , the host 102 , the host 103 and the host 104 are all connected to the same shared storage 105 . Each host can write its own time stamp in the corresponding position of the shared storage, and send the waiting time of the IO operation to the shared storage while writing the time stamp each time.

Please refer to FIG. 2 , which is an implementation flowchart of an IO operation method based on distributed lock storage in the present invention. This method can be applied to the first host, and the first host and multiple other hosts are connected to the same shared storage. The method may include the steps of:

S201: Read, according to a preset read cycle, the time stamp written to the shared storage by each other host according to a preset write cycle.

Shared storage can be one or a group of resources, and there can be multiple hosts connected to the same shared storage. Each host connected to the same shared storage can write its own timestamp to the shared storage. In this application document, the time stamp of the host refers to the current boot time of the host, for example, it may be the number of seconds of the current boot time of the host. Each host can start a process to write its own timestamp in the shared storage according to the preset write cycle. The timestamp written by each host in the shared storage can only be read by other hosts and cannot be modified. For example, each joining host can be assigned a unique ID corresponding to the host on the shared storage. space, each host can read and write to its own space, and can only read and write to the space corresponding to other hosts.

After the first host reads the time stamp written into the shared storage by each other host according to the preset read cycle according to the preset read cycle, the operation of step S202 may be performed.

It should be noted that the reading cycle and the writing cycle can be set and adjusted according to actual conditions, which will not affect the implementation of the present invention. Preferably, the read period is greater than or equal to the write period.

S202: Detect whether the target data is occupied by the second host when issuing a first IO operation instruction to the target data in the shared storage.

If yes, go to step S203. Wherein, the second host is any one of multiple other hosts.

The target data is the data stored in the shared storage. When the first host issues a first IO operation command to the target data, it is detected whether the target data is occupied by the second host. The second host and the first host are connected to the same shared storage, and when the target data is occupied by the second host, the first host cannot issue an IO operation command to the target data. For example, when the second host performs a write operation on the target data, an identifier is set on the target data through the distributed lock, and the identifier may carry the name of the second host. When the target data has the identifier of the second host, the first host determines that the target data is occupied by the second data.

Of course, if it is determined that the target data is not occupied by any other host, the first host may issue an IO operation command to the target data to complete corresponding read and write operations.

S203: Read the time stamp of the second host, and determine whether the second host is offline according to the time stamp of the second host read this time and the time stamp of the second host read last time.

After detecting that the target data is occupied by the second host, the first host reads the timestamp of the second host, and according to the timestamp of the second host read this time and the timestamp of the second host read last time, Determine if the secondary host is offline.

The first host can compare the timestamp of the second host read this time with the timestamp of the second host read last time to determine whether the timestamps of the two reads have changed. If the time of the two reads There is no change in the stamp, and it can be determined that the second host is offline.

Of course, in a specific implementation of the present invention, considering that the IO operations on the target data may be relatively frequent, or the preset reading cycle is relatively short, when the first host determines that the second host to read this time When the timestamp of the second host has not changed compared with the last read timestamp of the second host, it can be further determined whether the time interval between two read timestamps exceeds the write cycle of the second host. If not, it can be done in a After waiting for a certain amount of time, read the timestamp of the second host. If the read time stamp still does not change after the waiting period, it may be determined that the second host is offline.

For example, the preset writing period of the second host is 10 seconds, and the preset reading period of the first host is 30 seconds. Usually, the preset read cycle is greater than or equal to the preset write cycle. When the first host detects that the target data is occupied by the second host, the read timestamp of the second host is one hour, and the last read timestamp of the second host is also one hour, and the second host is read twice The timestamp interval is 7 seconds. Since 7 seconds is lower than the write period of the second host of 10 seconds, the timestamp of the second host can be read after a waiting period. The waiting time can be set and adjusted according to the actual situation, for example, it can be set to be the same as the writing cycle of the second host, or it can be set to be three times the writing cycle of the second host, etc., which is not limited by the present invention . For example, when the waiting period is set to 20 seconds, the timestamp of the second host is read after 20 seconds, and if the read timestamp of the second host is still one hour, it is determined that the second host is offline.

After the first host determines that the second host is offline, the operation of S204 may be performed.

S204: Release the occupation of the target data by the second host and perform an IO operation.

After determining that the second host is offline, the first host releases the occupation of the target data by the second host. For example, the identifier set by the second host on the target data through the distributed lock may be deleted. After the first host releases the occupation of the target data by the second host, it can perform an IO operation on the target data.

By applying the technical solution provided by the embodiment of the present invention, the time stamp written by each other host to the shared storage according to the preset write cycle is read according to the preset read cycle, and the target data in the shared storage is issued During the first IO operation instruction, detect whether the target data is occupied by the second host, if yes, read the timestamp of the second host, according to the timestamp of the second host read this time and the second read last time The time stamp of the host, to determine whether the second host is offline, and if so, release the occupation of the target data by the second host and perform IO operations.

When the reading and writing pressure of the shared storage is relatively high, it may not be able to respond to the IO operation of the second host in time. Therefore, if the IO operation command issued by the second host has an IO timeout, it may be that the second host has disconnected from the shared storage. The connection causes the IO timeout of the second host, or the IO pressure of the shared storage may cause the IO timeout of the second host. In the solution of the present invention, when the first host detects that the target data is occupied by the second host, it reads the time stamp of the second host, and compares the time stamp of the second host read this time with the time stamp of the second host read last time. The host's timestamp is compared to determine whether the second host is offline. When the IO pressure of the shared storage is high, the IO of the second host will time out, but the second host can send a time stamp to make the first host determine whether the second host is offline, avoiding that the second host is not offline, but due to The IO timeout of the second host causes unexpected isolation of the second host, so that the IO operation on the shared storage cannot be completed.

In a specific implementation manner of the present invention, step S203 includes:

Step 1: Obtain the IO waiting time of the second host;

Step 2: Determine whether the timestamp of the second host read last time plus the IO waiting time is greater than the timestamp of the second host read this time, if not, then enter step 3;

Step 3: Make sure that the second host is offline.

The first host reads the IO waiting time of the second host from the shared storage, and the IO waiting time of the second host is sent to a corresponding area of the shared storage by the second host.

It should be noted that the IO waiting time of the second host may be determined according to the waiting time of the second IO command queue of the second host. Wherein, the second IO instruction queue has a non-timeout attribute, and includes a second IO operation instruction issued by the second host to the target data. Specifically: the second host may add a second IO command queue with a non-timeout attribute, the queue is composed of second IO operation commands. The non-timeout attribute refers to that when the second host issues the IO second IO operation command in the queue, if the delivery fails, it will continue to issue, and the queue will not be deleted until the second IO operation command in the queue is issued. sent successfully. Of course, each host connected to the same shared storage can use the IO command queue with non-timeout property to issue IO operation commands.

The second host may deliver the IO waiting time of the second host while delivering the time stamp. Of course, other delivery schemes may also be adopted for the delivery of the IO waiting time, which does not affect the implementation of the present invention. The first host adds the timestamp of the second host read last time to the IO waiting time of the second host, and determines whether the addition result is greater than the timestamp of the second host read this time, and if not, then determines the first The second host is offline.

If only the timestamp of the second host read this time and the timestamp of the second host read last time are used to determine whether the second host is offline, then the read cycle of the first host is shorter, and the shared storage target Multiple factors such as frequent data IO operations and possible delays in sending timestamps by the second host will affect the accuracy of the first host in judging whether the second host is offline. When the second host has not disconnected from the shared storage, and the IO operation command of the second host times out due to the high IO pressure of the shared storage, the time stamp of the second host read last time plus the first The sum of the IO waiting time of the two hosts will be greater than or equal to the timestamp of the second host read this time. When the second host disconnects from the shared storage, for example, the second host crashes or the network connection of the second host is interrupted, because the second host cannot issue the IO waiting time to update the IO of the second host saved in the shared storage Waiting time, the first host will sum the IO waiting time of the second host read from the shared storage with the timestamp of the second host read last time, and the result will be less than the time stamp of the second host read this time timestamp.

For example, when the first host detects that the second host occupies the target data in the shared storage, the time stamp of the second host read this time is 60 seconds, the IO waiting time of the second host is 32 seconds, and the time stamp of the last read The timestamp of the second host is 30 seconds, then the time stamp of the second host read last time is added to the IO waiting time of the second host, and the time obtained is greater than the timestamp of the second host read this time, Then it can be determined that the second host has not disconnected from the shared storage.

Corresponding to the above method embodiment, the embodiment of the present invention also provides an IO operation device based on distributed lock storage, the IO operation device based on distributed lock storage described below is the same as the one described above The IO operation methods of storage based on distributed locks can be referred to each other.

Referring to Figure 3, it is a schematic structural diagram of an IO operation device based on distributed lock storage in an embodiment of the present invention, which is applied to the first host, and the first host and multiple other hosts are connected to the same shared storage. The device includes the following modules:

The timestamp reading module 301 is configured to read the timestamp written by each other host to the shared storage according to the preset write cycle according to the preset read cycle;

The target data detection module 302 is configured to detect whether the target data is occupied by the second host when issuing the first IO operation instruction to the target data in the shared storage, and if so, trigger the second host offline determination module 303, the second The host is any one of multiple other hosts;

The second host offline determination module 303 is configured to read the timestamp of the second host, and determine whether the second host is offline according to the timestamp of the second host read this time and the timestamp of the second host read last time , if yes, trigger the IO operation module 304;

The IO operation module 304 is configured to release the occupation of the target data by the second host and perform an IO operation.

Apply the device provided by the embodiment of the present invention to read the time stamps written by each other host to the shared storage according to the preset read cycle according to the preset read cycle, and send the second time stamp to the target data in the shared storage During an IO operation command, detect whether the target data is occupied by the second host, if yes, read the timestamp of the second host, according to the timestamp of the second host read this time and the second host read last time to determine whether the second host is offline, and if so, release the second host from occupying the target data and perform IO operations.

When the reading and writing pressure of the shared storage is relatively high, it may not be able to respond to the IO operation of the second host in time. Therefore, if the IO operation command issued by the second host has an IO timeout, it may be that the second host has disconnected from the shared storage. The connection causes the IO timeout of the second host, or the IO pressure of the shared storage may cause the IO timeout of the second host. In the solution of the present invention, when the first host detects that the target data is occupied by the second host, it reads the time stamp of the second host, and compares the time stamp of the second host read this time with the time stamp of the second host read last time. The host's timestamp is compared to determine whether the second host is offline. When the IO pressure of the shared storage is high, the IO of the second host will time out, but the second host can send a time stamp to make the first host determine whether the second host is offline, avoiding that the second host is not offline, but due to The IO operation of the second host fails to complete the IO operation on the shared storage due to timeout.

In a specific implementation manner of the present invention, the module 303 for determining that the second host is offline includes:

The IO waiting time obtaining submodule is used to obtain the IO waiting time of the second host;

Timestamp determines submodule, is used to determine whether the timestamp of the second host read last time plus the IO waiting time is greater than the timestamp of the second host read this time, if not, then enters the offline determination submodule of the second host module;

The second host offline determining submodule is configured to determine that the second host is offline.

In a specific embodiment of the present invention, the IO waiting time acquisition submodule is specifically used for:

Determine the IO waiting time of the second host according to the waiting time of the second IO command queue of the second host;

The second IO instruction queue has a non-timeout attribute, and includes a second IO operation instruction issued by the second host to the target data.

In a specific implementation manner of the present invention, the first IO operation instruction is an instruction in the first IO instruction queue with a non-timeout attribute.

In a specific embodiment of the present invention, the read period is greater than or equal to the write period.

Each embodiment in this specification is described in a progressive manner, each embodiment focuses on the difference from other embodiments, and the same or similar parts of each embodiment can be referred to each other. As for the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and for the related information, please refer to the description of the method part.

Professionals can further realize that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, computer software or a combination of the two. In order to clearly illustrate the possible For interchangeability, in the above description, the composition and steps of each example have been generally described according to their functions. Whether these functions are executed by hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art may use different methods to implement the described functions for each specific application, but such implementation should not be regarded as exceeding the scope of the present invention.

The steps of the methods or algorithms described in connection with the embodiments disclosed herein may be directly implemented by hardware, software modules executed by a processor, or a combination of both. Software modules can be placed in random access memory (RAM), internal memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other Any other known storage medium.

In this paper, specific examples are used to illustrate the principles and implementation methods of the present invention, and the descriptions of the above embodiments are only used to help understand the technical solutions and core ideas of the present invention. It should be pointed out that for those skilled in the art, without departing from the principles of the present invention, some improvements and modifications can be made to the present invention, and these improvements and modifications also fall within the protection scope of the claims of the present invention.

Claims (10)

1. A stored IO operation method based on a distributed lock, characterized in that it is applied to a first host, and the first host is connected to the same shared storage with multiple other hosts, including: reading the time stamp written by each other host to the shared storage according to the preset write cycle according to the preset read cycle; When issuing a first IO operation instruction to the target data in the shared storage, detecting whether the target data is occupied by a second host, where the second host is any one of a plurality of other hosts; If yes, read the timestamp of the second host, and determine the second host according to the timestamp of the second host read this time and the timestamp of the second host read last time whether offline; If so, release the occupation of the target data by the second host and perform an IO operation. 2. The method according to claim 1, wherein the second host is determined according to the timestamp of the second host read this time and the timestamp of the second host read last time. Whether the second host is offline, including: Obtain the IO waiting time of the second host; Determine whether the timestamp of the second host read last time plus the IO waiting time is greater than the timestamp of the second host read this time; If not, it is determined that the second host is offline. 3. The method according to claim 2, wherein the obtaining the IO waiting time of the second host comprises: determining the IO waiting time of the second host according to the waiting time of the second IO command queue of the second host; Wherein, the second IO instruction queue has a non-timeout attribute, and includes a second IO operation instruction issued by the second host to the target data. 4. The method according to claim 1, wherein the first IO operation instruction is an instruction in the first IO instruction queue with a non-timeout attribute. 5. The method according to any one of claims 1 to 4, wherein the read period is greater than or equal to the write period. 6. An IO operation device based on distributed lock storage, characterized in that it is applied to a first host, and the first host is connected to the same shared storage with multiple other hosts, including: A time stamp reading module, configured to read the time stamp written by each other host to the shared storage according to the preset write cycle according to the preset read cycle; The target data detection module is used to detect whether the target data is occupied by the second host when issuing the first IO operation instruction to the target data in the shared storage, and if so, trigger the second host offline determination module, The second host is any one of multiple other hosts; The second host offline determination module is configured to read the timestamp of the second host, according to the timestamp of the second host read this time and the timestamp of the second host read last time , determining whether the second host is offline, and if so, triggering the IO operation module; The IO operation module is configured to release the occupation of the target data by the second host and perform an IO operation. 7. The device according to claim 6, wherein the module for determining that the second host is offline comprises: The IO waiting time obtaining submodule is used to obtain the IO waiting time of the second host; The timestamp determination submodule is used to determine whether the timestamp of the second host read last time plus the IO waiting time is greater than the timestamp of the second host read this time, if not, enter The second host determines the submodule offline; The second host offline determination submodule is configured to determine that the second host is offline. 8. The device according to claim 7, wherein the IO waiting time obtaining submodule is specifically used for: determining the IO waiting time of the second host according to the waiting time of the second IO command queue of the second host; Wherein, the second IO instruction queue has a non-timeout attribute, and includes a second IO operation instruction issued by the second host to the target data. 9. The device according to claim 6, wherein the first IO operation instruction is an instruction in the first IO instruction queue with a non-timeout attribute. 10. The device according to any one of claims 6 to 9, wherein the read cycle is greater than or equal to the write cycle.