CN116955013A - Method, system and digital networking system for recovering memory state data for data language - Google Patents
Method, system and digital networking system for recovering memory state data for data language Download PDFInfo
- Publication number
- CN116955013A CN116955013A CN202310968475.9A CN202310968475A CN116955013A CN 116955013 A CN116955013 A CN 116955013A CN 202310968475 A CN202310968475 A CN 202310968475A CN 116955013 A CN116955013 A CN 116955013A
- Authority
- CN
- China
- Prior art keywords
- digital object
- object warehouse
- node
- data
- warehouse node
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 89
- 230000006855 networking Effects 0.000 title claims abstract description 26
- 238000011084 recovery Methods 0.000 claims description 113
- 238000004458 analytical method Methods 0.000 claims description 19
- 230000002688 persistence Effects 0.000 claims description 12
- 230000015556 catabolic process Effects 0.000 abstract description 3
- 238000012360 testing method Methods 0.000 description 33
- 230000001934 delay Effects 0.000 description 10
- 230000008569 process Effects 0.000 description 10
- 238000010586 diagram Methods 0.000 description 8
- 230000009471 action Effects 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 5
- 230000008859 change Effects 0.000 description 5
- 230000007246 mechanism Effects 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 238000004590 computer program Methods 0.000 description 3
- 230000000977 initiatory effect Effects 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 229920004934 Dacron® Polymers 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 229910003460 diamond Inorganic materials 0.000 description 1
- 239000010432 diamond Substances 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000013467 fragmentation Methods 0.000 description 1
- 238000006062 fragmentation reaction Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000002045 lasting effect Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000002085 persistent effect Effects 0.000 description 1
- 239000005020 polyethylene terephthalate Substances 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
- G06F11/14—Error detection or correction of the data by redundancy in operation
- G06F11/1402—Saving, restoring, recovering or retrying
- G06F11/1446—Point-in-time backing up or restoration of persistent data
- G06F11/1448—Management of the data involved in backup or backup restore
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
- G06F11/14—Error detection or correction of the data by redundancy in operation
- G06F11/1402—Saving, restoring, recovering or retrying
- G06F11/1446—Point-in-time backing up or restoration of persistent data
- G06F11/1458—Management of the backup or restore process
- G06F11/1469—Backup restoration techniques
Abstract
The invention discloses a method and a system for recovering memory state data for a data language and a digital networking system, which relate to the field of digital networking and comprise the following steps: the master digital object warehouse node receives and executes the call request, and simultaneously broadcasts the call request to all slave digital object warehouse nodes; if the calling result of any slave digital object warehouse node is wrong, the master digital object warehouse node sends memory state data to the slave digital object warehouse node, and if the calling result of the master digital object warehouse node is wrong, the memory state data is acquired from any correct slave digital object warehouse node. The invention ensures that the memory state data for the data language can be automatically recovered after the problems of breakdown, downtime, power failure and the like of the digital object of the computer node occur and the normal state is recovered in the operation life cycle of the multi-point data language.
Description
Technical Field
The invention relates to the field of digital networking, in particular to a method and a system for recovering memory state data for digital networking data languages and a digital networking system.
Background
With the advent of the big data age, the value of data is increasingly prominent. Through analysis of the big data, many additional values can be generated. For example, during an epidemic situation, through big data analysis, a person can accurately and quickly inquire about which cities to go, which trains to take, whether to go to a high risk area, and the like. To use data for analysis, exchange and processing of data is required. In a data sharing exchange scenario, there is a regulatory problem, i.e., a serious trust problem, on large-scale shared data. The acquirer needs to use the acquirer's data, producing valuable results by manipulating the data, but the acquirer is often not trusted by the acquirer. Once the data is transferred from the supplier to the acquirer, the acquirer can do any action, such as doing something different from the previously declared action on the data, even vending the original data.
Based on the above-mentioned big data application requirements, there is a big data digital networking system for running data languages based on memory (the data languages refer to a digital object, the data languages are divided into general data language uses and cluster data language uses, the general data language uses need to be created and executed in one data language use warehouse, and the cluster data language uses need to be created and executed in a plurality of data language use warehouses). Corresponding services are provided for scenes of big data application. In the big data networking system adopting the memory computing mode for the data language, the sharing exchange of big data is realized by means of the form of the data language, and in the big data networking system adopting the memory computing mode for the data language, the memory-based computing mode and the random multipoint executing mode are adopted for the data language.
The memory-based calculation mode can well meet the requirement of the execution time of the data language, improves the throughput rate of the execution of the data language, has good overall performance and has no additional cost. However, based on a memory-based computing mode, as the data words are used in the memory for computing, running and storing, when unreliability occurs, the devices running the data words are in a plurality of points (Beijing university points are ubiquitous interoperation platforms developed by Beijing university system software teams for man-machine object fusion computing scenes, so that the problems of on-demand interoperation and trusted management of data resources of ubiquitous devices such as cloud, network, edge, end and the like) digital object collapse, downtime, power failure and the like are solved, and the memory state data of the whole data words can be completely lost and cannot be recovered. Therefore, in the memory-based computing mode, on the basis of ensuring the execution time requirement and the execution throughput rate of the data language, the problem that the data language memory state data automatically recovers after the related equipment has the problems of crashing, downtime, power failure and the like of the digital object is guaranteed, and the data language memory state data can be recovered automatically is a problem to be solved urgently.
Disclosure of Invention
In view of the above problems, the invention provides a method, a device and a digital networking system for recovering memory state data for a data language, which ensure that the memory state data for the data language can automatically recover the memory state data for the data language after the problems of crashing, downtime, power failure and the like of a digital object of the related equipment and recovering the memory state data for the data language are solved on the basis of guaranteeing the execution time requirement for the data language and the execution throughput rate for the data language.
The embodiment of the invention provides a recovery method of memory state data for a data language, which is applied to a plurality of digital object warehouse nodes of a digital network, each digital object warehouse node of the plurality of digital object warehouse nodes of the digital network is operated with the data language, the data language is a multi-point data language, the plurality of digital object warehouse nodes of the digital network form a data language cluster, and the recovery method comprises the following steps:
receiving and executing a call request by a master number and rayleigh number digital object warehouse node in the data language cluster, and broadcasting the call request to all slave number and rayleigh number digital object warehouse nodes in the data language cluster by the master number and rayleigh number digital object warehouse node;
The main digital object warehouse node receives all calling results fed back by the secondary digital object warehouse node;
under the condition that the calling result fed back by a first slave digital object warehouse node in all slave digital object warehouse nodes is wrong, the master digital object warehouse node sends full-quantity state data used by multi-point data languages running by the master digital object warehouse node to the first slave digital object warehouse node;
under the condition that the master digital object warehouse node executes the calling request and determines that the calling result is wrong, the master digital object warehouse node sends a state data request to a second slave digital object warehouse node in all slave digital object warehouse nodes, wherein the second slave digital object warehouse node is any slave digital object warehouse node in all slave digital object warehouse nodes with correct calling results;
the master digital object warehouse node receives full-quantity state data for multi-point data languages running on the second slave digital object warehouse node, which is sent by the second slave digital object warehouse node;
In the case that the first slave digital object warehouse node is offline and online again, the first slave digital object warehouse node sends recovery information to the master digital object warehouse node, and the master digital object warehouse node sends full state data for the self-operated multi-point data language to the first slave digital object warehouse node or sends incremental state data for the self-operated multi-point data language to the first slave digital object warehouse node or does not send any data to the first slave digital object warehouse node according to the recovery information and the current operation mode of the data language cluster;
when the master number Rayleigh-end digital object warehouse node is offline and online again, and the master number Rayleigh-end digital object warehouse node is changed into a slave number Rayleigh-end digital object warehouse node, and the second slave number Rayleigh-end digital object warehouse node is changed into a new master number Rayleigh-end digital object warehouse node, the master number Rayleigh-end digital object warehouse node sends the recovery information to the second slave number Rayleigh-end digital object warehouse node, and the second slave number Rayleigh-end digital object warehouse node sends full state data for self-running multi-point data words to the master number Rayleigh-end digital object warehouse node or sends incremental state data for self-running multi-point data words to the master number Rayleigh-end digital object warehouse node or does not send any data to the master number Rayleigh-end digital object warehouse node according to the recovery information and the current running mode of the data words;
And under the condition that the main number rayleigh operator digital object warehouse node is offline and online again and the main number rayleigh operator digital object warehouse node is still the main number rayleigh operator digital object warehouse node, the main number rayleigh operator digital object warehouse node executes the following steps: and receiving and executing a call request, wherein the master digital object warehouse node broadcasts the call request to all slave digital object warehouse nodes in the data language cluster.
Optionally, before the master number rake digital object repository node in the data language cluster receives and executes the call request, the method further includes:
the main number and rayleigh operator digital object warehouse node determines that the operation mode is a common operation mode under the condition that the multi-point data language is classified into stateless data language;
the main number-end operator digital object warehouse node is classified into a stateful data language in the multi-point data language, and the main number-end operator digital object warehouse node determines the operation mode to be the common operation mode under the condition that the first preset condition is met between the total number of the number-end operator digital object warehouse nodes in the data language cluster and the total number of the number-end operator digital object warehouse nodes in the current normal operation;
The main digital data object warehouse node is classified into a stateful data language in the multi-point data language, and the main digital data object warehouse node determines the operation mode to be the stable operation mode under the condition that a second preset condition is met between the total number of the digital data object warehouse nodes of the digital data language cluster and the total number of the digital data object warehouse nodes of the digital data operator in the current normal operation;
under the condition of the common operation mode, each data-language cluster operates in a memory persistence mode by each data-language operator digital object warehouse node;
under the condition of the stable operation mode, each digital object warehouse node records the full-quantity state data and the increment state data for the multi-point data language into a hard disk of the node;
when the recording quantity of the incremental state data for the multi-point data language reaches a preset quantity, each digital object warehouse node of the digital object database of the data Rate respectively updates and records the full state data for the multi-point data language, and simultaneously respectively deletes the incremental state data records for the multi-point data language;
wherein, the first preset condition is: the total number of the current normal running digital object warehouse nodes of the digital speech clusters is larger than the total number of the preset operation on the total number of the digital object warehouse nodes of the digital speech clusters, and the preset operation is as follows: half the total number of the digital object warehouse nodes of the number-of-Rayleigh operator in the data language cluster to obtain a first result, and rounding the first result upwards;
The second preset condition is: the total number of the digital object warehouse nodes of the number Rayleigh operator which is currently and normally operated, and the total number is not larger than the total number after the preset operation is carried out on the total number of the digital object warehouse nodes of the number Rayleigh operator in the data language cluster.
Optionally, the recovery information includes: the operation mode of the data language cluster before the first slave digital object warehouse node is offline and the version number of the multi-point data language operated after the first slave digital object warehouse node is offline again;
in the case that the first slave digital object warehouse node is offline and online again, the first slave digital object warehouse node transmits restoration information to the master digital object warehouse node, and the master digital object warehouse node transmits full state data for a multi-point data language running by itself to the first slave digital object warehouse node or transmits incremental state data for the multi-point data language running by itself to the first slave digital object warehouse node or does not transmit any data to the first slave digital object warehouse node according to the restoration information and the current operation mode of the data language cluster, including:
When the data language cluster is in the normal operation mode before the first slave digital object warehouse node is offline, and the data language cluster is in any operation mode after the first slave digital object warehouse node is online again, the master digital object warehouse node determines whether the version number of the multi-point data language cluster operated after the first slave digital object warehouse node is online again is the same as the current state version number of the data language cluster, if the version number of the multi-point data language cluster is the same, the master digital object warehouse node does not send any data to the first slave digital object warehouse node, if the version number of the multi-point data language cluster is not the same, the master digital object warehouse node sends the full state data of the multi-point data language operated by the master digital object warehouse node to the first slave digital object warehouse node, so that the first slave digital object warehouse node operates the multi-point data language of the master digital object warehouse node based on the received full state data;
when the data language cluster is in the stable operation mode before the first slave digital object warehouse node is offline and the data language cluster is in the normal operation mode after the first slave digital object warehouse node is offline again, the master digital object warehouse node sends full-quantity state data of the self-operated multi-point data language to the first slave digital object warehouse node so that the first slave digital object warehouse node operates the self multi-point data language based on the received full-quantity state data;
Under the condition that the data word cluster is in the stable operation mode before the first slave digital object warehouse node is offline, and the data word cluster is in the stable operation mode after the first slave digital object warehouse node is offline again, the first slave digital object warehouse node acquires full-quantity state data and incremental state data for the multi-point data word from a hard disk of the first slave digital object warehouse node, meanwhile, the master digital object warehouse node determines whether a version number for the multi-point data word operated after the first slave digital object warehouse node is offline again is the same as a current state version number of the data word cluster, if the version number is the same, the master digital object warehouse node does not send any data to the first slave digital object warehouse node, and if the version number is different, the master digital object warehouse node sends the new incremental state data for the multi-point data word operated by the master digital object warehouse node to the first slave digital object warehouse node, so that the first slave digital object warehouse node does not receive the full-quantity state data word incremental data for the multi-point data word based on the first slave digital object warehouse node and the full-quantity data word cluster, and the incremental state data from the first slave digital object warehouse node does not receive the full-quantity data word from the first slave digital object warehouse node.
Optionally, the recovery information further includes: the operation mode of the data language cluster before the main number and the plurality of data language version numbers are used for the operation after the main number and the plurality of data object nodes are used for the operation;
in the case where the master number rayleigh digital object warehouse node is offline and online again, and the master number rayleigh digital object warehouse node becomes a slave number rayleigh digital object warehouse node, the master number rayleigh digital object warehouse node transmits the restoration information to the second slave number rayleigh digital object warehouse node, and the second slave number rayleigh digital object warehouse node transmits full-scale state data for a self-operated multi-point data word to the master number rayleigh digital object warehouse node or transmits incremental state data for a self-operated multi-point data word to the master number rayleigh digital object warehouse node or does not transmit any data to the master number rayleigh digital object warehouse node according to the restoration information and the current operation mode of the data word cluster, including:
Under the condition that the data language cluster is in the common operation mode before the main number and the plurality of sub-operator digital object warehouse nodes are off line and the data language cluster is in any operation mode after the main number and the sub-operator digital object warehouse nodes are on line again, the second slave digital object warehouse node determines whether the version number of the multi-point data language running after the master digital object warehouse node is on line again is the same as the version number of the current state of the data language cluster, if the data are the same, the second slave digital object warehouse node does not send any data to the master digital object warehouse node, and if the data are not the same, the second slave digital object warehouse node sends the total state data for the self-operated multi-point data language to the master digital object warehouse node, so that the master digital object warehouse node operates the self-operated multi-point data language based on the received total state data;
when the data language cluster is in the stable operation mode before the master number and Rayleigh operator digital object warehouse node is offline and the data language cluster is in the normal operation mode after the master number and Rayleigh operator digital object warehouse node is offline again, the second slave number and Rayleigh operator digital object warehouse node sends full-quantity state data of the self-operated multi-point data language to the master number and Rayleigh operator digital object warehouse node so that the master number and Rayleigh operator digital object warehouse node operates the self multi-point data language based on the received full-quantity state data;
And under the condition that the data word cluster is in the stable operation mode before the master number Rayleigh operator digital object warehouse node is offline and the data word cluster is in the stable operation mode after the master number Rayleigh operator digital object warehouse node is offline again, the master number Rayleigh operator digital object warehouse node acquires full-quantity state data and incremental state data for the multi-point data word from a hard disk of the master number Rayleigh operator digital object warehouse node, and meanwhile, the second slave number Rayleigh operator digital object warehouse node determines whether the version number for the multi-point data word operated after the master number Rayleigh operator digital object warehouse node is offline again is the same as the current state version number of the data word cluster, if the version number for the multi-point data word is the same, the second slave number Rayleigh operator digital object warehouse node does not send any data to the master number Rayleigh operator digital object warehouse node, and if the version number for the multi-point data word operated by the second slave number Rayleigh operator digital object warehouse node is different, the second slave number Rayleigh operator digital object warehouse node sends the new incremental state data for the multi-point data word operated by the master number Rayleigh operator digital object warehouse node to the master number warehouse node so that the master number Rayleigh operator digital object warehouse node obtains the full-quantity data based on the received from the state data of the master number Rayleigh operator digital object warehouse node and the full-quantity data.
Optionally, the master digital object repository node determines whether the version number for the multi-point data language operated after the first slave digital object repository node is on line again is the same as the current state version number of the cluster for the data language, including:
under the condition that a multi-point data language version number used by the first slave digital object warehouse node is not crashed, the recovery information sent to the master digital object warehouse node by the first slave digital object warehouse node carries the multi-point data language version number operated after the first slave digital object warehouse node is on line again;
the master digital object warehouse node determines whether the version number for the multi-point data language operated after the first slave digital object warehouse node is on line again is the same as the current state version number of the data language cluster or not based on the received version number for the multi-point data language operated after the first slave digital object warehouse node is on line again;
under the condition that a multi-point data language version number on the first slave digital object warehouse node breaks down, the recovery information sent to the master digital object warehouse node by the first slave digital object warehouse node does not carry the multi-point data language version number operated after the first slave digital object warehouse node is on line again;
And the master digital object warehouse node does not receive the version number for the multi-point data language operated after the first slave digital object warehouse node is on line again, and the version number for the multi-point data language operated after the first slave digital object warehouse node is on line again is determined to be different from the current state version number of the data language cluster.
Optionally, the recovery method further includes:
before the first slave digital object warehouse node is offline, if the total number of the digital object warehouse nodes of the data language cluster is between the total number of the digital object warehouse nodes of the digital object and the total number of the digital object warehouse nodes of the digital object in the current normal operation, the master digital object warehouse node determines that the operation mode is the common operation mode;
after the first slave digital object warehouse node is offline, and when the first slave digital object warehouse node is not online again, if the total number of digital object warehouse nodes of the digital language cluster is between the total number of digital object warehouse nodes of the first slave digital object warehouse node and the total number of digital object warehouse nodes of the current normal operation, the master digital object warehouse node still meets the first preset condition, and the operation mode is still the normal operation mode; or alternatively
If the second preset condition is met between the total number of the digital object warehouse nodes of the number Rayleigh operator in the data language cluster and the total number of the digital object warehouse nodes of the number Rayleigh operator in the current normal operation, the main digital object warehouse node switches the operation mode into the stable operation mode;
when the first slave digital object warehouse node is offline and is not online again, and the operation mode is the normal operation mode, after the first slave digital object warehouse node is online again, if the total number of the digital object warehouse nodes of the data language cluster is between the total number of the digital object warehouse nodes of the current normal operation and the total number of the digital object warehouse nodes of the current normal operation, the master digital object warehouse node keeps the operation mode to be the normal operation mode; or alternatively
If the second preset condition is met between the total number of the digital object warehouse nodes of the number Rayleigh operator in the data language cluster and the total number of the digital object warehouse nodes of the number Rayleigh operator in the current normal operation, the main digital object warehouse node switches the operation mode into the stable operation mode;
When the first slave digital object warehouse node is offline and is not online again, and the operation mode is the stable operation mode, after the first slave digital object warehouse node is online again, if the total number of the digital object warehouse nodes of the data language cluster is between the total number of the digital object warehouse nodes of the current normal operation and the total number of the digital object warehouse nodes of the current normal operation, the master digital object warehouse node switches the operation mode into the normal operation mode; or alternatively
If the second preset condition is still met between the total number of the digital object warehouse nodes of the number of the Rayleigh operators in the data language cluster and the total number of the digital object warehouse nodes of the number Rayleigh operators in the current normal operation, the main digital object warehouse node keeps the operation mode still to be the stable operation mode; after the master number and Rayleigh operator digital object warehouse node switches the operation mode from the stable operation mode to the normal operation mode, all the number and Rayleigh operator digital object warehouse nodes in the data language cluster delete the full-quantity state data and the increment state data recorded in the hard disk during the stable operation mode.
Optionally, the master number rake digital object warehouse node is offline and online again, and the master number rake digital object warehouse node becomes a slave number rake digital object warehouse node, and the second slave number rake digital object warehouse node becomes a master number rake digital object warehouse node, including:
under the condition that the main digital data object warehouse node is offline and the offline time length exceeds the preset time length and is not online again, other digital data object warehouse nodes in the data language cluster are disconnected with the main digital data object warehouse node, and a reelect request is sent to a digital data networking identification analysis system in the digital network;
after receiving re-election requests sent by all the digital object warehouse nodes except the main digital object warehouse node, the digital networking identification analysis system elects the second slave digital object warehouse node from all the digital object warehouse nodes except the main digital object warehouse node as the new main digital object warehouse node;
After the second slave digital object warehouse node becomes the new master digital object warehouse node, sending a connection request to all digital object warehouse nodes except the master digital object warehouse node so that all digital object warehouse nodes except the master digital object warehouse node are connected with the second slave digital object warehouse node;
and after the master digital object warehouse node is on line again, the second slave digital object warehouse node sends the connection request to the master digital object warehouse node so that the master digital object warehouse node connects the second slave digital object warehouse node with the identity of the slave digital object warehouse node.
Optionally, the primary number of rayleigh digital object warehouse nodes are offline and online again, and the primary number of rayleigh digital object warehouse nodes are still primary number of rayleigh digital object warehouse nodes, including:
under the condition that the main number Rayleigh operator digital object warehouse node is offline and the offline time is offline again when the preset time is not exceeded, other number Rayleigh operator digital object warehouse nodes in the data language cluster cannot sense that the main number Rayleigh operator digital object warehouse node is offline, and still remain connected with the main number Rayleigh operator digital object warehouse node;
And after the primary digital object warehouse node is online again, executing the following steps: and receiving and executing a call request, wherein the master digital object warehouse node broadcasts the call request to all slave digital object warehouse nodes in the data language cluster.
The embodiment of the invention also provides a recovery system of the memory state data for the data language, which comprises the following steps: a plurality of digital object repository nodes of a digital network, the plurality of digital object repository nodes of the digital network comprising: a master number of rayleigh digital object warehouse nodes and a slave number of rayleigh digital object warehouse nodes; each of the plurality of digital object warehouse nodes operates a data language, the data language is a multi-point data language, the plurality of digital object warehouse nodes form a cluster for the data language, and the main digital object warehouse node comprises:
the receiving and executing broadcasting module is used for receiving and executing the calling request and broadcasting the calling request to all slave digital object warehouse nodes in the data language cluster;
The receiving result module is used for receiving all calling results fed back from the digital object warehouse node of the digital Rayleigh operator;
the first sending module is used for sending the full-quantity state data of the multi-point data language running by the first slave digital object warehouse node to the first slave digital object warehouse node under the condition that the calling result fed back by the first slave digital object warehouse node in all slave digital object warehouse nodes is wrong;
the data request module is used for sending a state data request to a second slave digital object warehouse node in all slave digital object warehouse nodes under the condition that the calling request is executed and the calling result of the data request module is determined to be wrong, wherein the second slave digital object warehouse node is any slave digital object warehouse node in all slave digital object warehouse nodes with correct calling results;
a data receiving module, configured to receive full-quantity state data for a multi-point data language running on the second slave digital object warehouse node, where the full-quantity state data is sent by the second slave digital object warehouse node;
the receiving information sending data module is used for receiving recovery information sent by the first slave digital object warehouse node when the first slave digital object warehouse node is offline and online again, sending full state data for the self-operated multi-point data language to the first slave digital object warehouse node or sending the incremental state data for the self-operated multi-point data language to the first slave digital object warehouse node or not sending any data to the first slave digital object warehouse node according to the recovery information and the current operation mode of the data language cluster;
The information sending and receiving data module is used for sending the recovery information to the second slave digital object warehouse node under the condition that the second slave digital object warehouse node is offline and online again and becomes a slave digital object warehouse node, and the second slave digital object warehouse node becomes a new master digital object warehouse node, and receiving the total state data or the incremental state data for the multi-point data words operated by the second slave digital object warehouse node according to the recovery information and the operation mode;
the execution module is used for executing the steps under the condition that the execution module is offline and online again and is still a master number and Rate operator digital object warehouse node: and receiving and executing the call request, and broadcasting the call request to all slave digital object warehouse nodes in the data language cluster.
Optionally, the primary number rake digital object warehouse node further includes: a determine operation mode module for:
under the condition that the classification of the multi-point data language is a stateless data language, determining that the operation mode is a common operation mode;
Determining that the operation mode is the common operation mode when the classification of the multi-point data language is a stateful data language and the total number of the digital object warehouse nodes of the number Rayleigh operator in the data language cluster is between the total number of the digital object warehouse nodes of the number Rayleigh operator in the current normal operation and the total number of the digital object warehouse nodes of the number Rayleigh operator in the current normal operation meets a first preset condition;
determining that the operation mode is the stable operation mode under the condition that the classification of the multi-point data language is used as a stateful data language, and the second preset condition is met between the total number of the digital object warehouse nodes of the number of the Rayleigh operators in the data language cluster and the total number of the digital object warehouse nodes of the number of the Rayleigh operators in the current normal operation;
the master digital data object warehouse node and the slave digital data object warehouse node each comprise: the operation record module is used for:
under the condition of the common operation mode, operating in a memory persistence mode;
under the condition of the stable operation mode, recording the full-quantity state data and the increment state data for the multi-point data language into a hard disk of the device;
when the recording quantity of the incremental state data for the multi-point data language reaches a preset quantity, respectively updating and recording the full state data for the multi-point data language, and respectively deleting the incremental state data records for the multi-point data language;
Wherein, the first preset condition is: the total number of the current normal running digital object warehouse nodes of the digital speech clusters is larger than the total number of the preset operation on the total number of the digital object warehouse nodes of the digital speech clusters, and the preset operation is as follows: half the total number of the digital object warehouse nodes of the number-of-Rayleigh operator in the data language cluster to obtain a first result, and rounding the first result upwards;
the second preset condition is: the total number of the digital object warehouse nodes of the number Rayleigh operator which is currently and normally operated, and the total number is not larger than the total number after the preset operation is carried out on the total number of the digital object warehouse nodes of the number Rayleigh operator in the data language cluster.
Optionally, the recovery information includes: the operation mode of the data language cluster before the first slave digital object warehouse node is offline and the version number of the multi-point data language operated after the first slave digital object warehouse node is offline again;
the information receiving and data transmitting module is specifically configured to:
when the data language cluster is in the normal operation mode before the first slave digital object warehouse node is offline, and the data language cluster is in any operation mode after the first slave digital object warehouse node is online again, determining whether the version number of the multi-point data language cluster operated after the first slave digital object warehouse node is online again is the same as the current state version number of the data language cluster, if so, not sending any data to the first slave digital object warehouse node, otherwise, sending the full state data of the multi-point data language operated by the first slave digital object warehouse node to the first slave digital object warehouse node, so that the first slave digital object warehouse node operates the multi-point data language of the first slave digital object warehouse node based on the received full state data;
When the data language cluster is in the stable operation mode before the first slave digital object warehouse node is offline and the data language cluster is in the normal operation mode after the first slave digital object warehouse node is online again, transmitting full-quantity state data of the multi-point data language running by the first slave digital object warehouse node to the first slave digital object warehouse node so that the first slave digital object warehouse node runs the multi-point data language by the first slave digital object warehouse node based on the received full-quantity state data;
and under the condition that the data word cluster is in the stable operation mode before the first slave digital object warehouse node is offline, and the data word cluster is in the stable operation mode after the first slave digital object warehouse node is offline again, the first slave digital object warehouse node acquires full-quantity state data and incremental state data for the multi-point data word from a hard disk of the first slave digital object warehouse node, simultaneously determines whether the version number for the multi-point data word running after the first slave digital object warehouse node is offline again is the same as the current state version number of the data word cluster, if the version number for the multi-point data word is the same as the current state version number of the data word cluster, any data is not transmitted to the first slave digital object warehouse node, and if the version number for the multi-point data word running by the first slave digital object warehouse node is different, the first slave digital object warehouse node is enabled to be based on the received new incremental state data and the full-quantity state data and the state data acquired from the hard disk of the first slave digital object warehouse node, and the first slave digital object warehouse node is not enabled to be the incremental state data of the multi-point data word from the hard disk of the first slave digital object warehouse node.
Optionally, the recovery information further includes: the operation mode of the data language cluster before the main number and the plurality of data language version numbers are used for the operation after the main number and the plurality of data object nodes are used for the operation;
the information sending and receiving data module is specifically configured to:
transmitting said recovery information to said second slave digital object warehouse node;
receiving the full state data of the multi-point data language transmitted by the second slave digital object warehouse node of the number and operating the multi-point data language of the slave digital object warehouse node based on the received full state data;
or acquiring full state data and incremental state data for the multi-point data language from the own hard disk, receiving the newly-added incremental state data for the multi-point data language sent by the second slave digital object warehouse node, and running the own multi-point data language based on the received newly-added incremental state data and the full state data and the incremental state data acquired from the own hard disk.
The embodiment of the invention also provides a digital networking system, which comprises: the system comprises a digital Rayleigh network identification analysis system, a plurality of digital object warehouse nodes stored in a digital manner and a plurality of digital object warehouse nodes stored in a digital manner;
The digital network identification analysis system is used for managing whether the digital object warehouse node of the digital sub-unit and the digital object warehouse node of the digital storage sub-digital sub-unit join the digital network system;
the plurality of digital object warehouse nodes are used for executing the recovery method of the memory state data for the data language according to any one of the above;
the plurality of digital object repository nodes are configured to store at least one type of information including:
the operation result of the multi-point data language;
state data for the multi-point data language;
and the execution branch information for the multi-point data language.
The invention provides a method for recovering memory state data for data languages, which is implemented by a main number and Rate operator digital object warehouse node receiving a call request, simultaneously broadcasting a call request to all slave digital object warehouse nodes; then receiving all calling results fed back from the digital object warehouse nodes of the digital Rayleigh operator; and when the calling result fed back by the first slave digital object warehouse node is wrong, namely, the data language of the multi-point data language running on the first slave digital object warehouse node is crashed, or the calling sequence number of the cross-data language is inconsistent with other digital object warehouse nodes, or the current state version number of the multi-point data language is inconsistent with other digital object warehouse nodes, the master digital object warehouse node sends the full state data of the multi-point data language running on the master digital object warehouse node to the first slave digital object warehouse node, so that the memory state data of the multi-point data language running on the first slave digital object warehouse node is recovered.
And in the case that the master digital object repository node executes the call request and determines that the call result is wrong, that is, the data language for the multi-point data language running on the master digital object repository node crashes, or the serial number of the cross-data language call is inconsistent with other digital object repository nodes, or the current state version number of the multi-point data language is inconsistent with other digital object repository nodes, the master digital object repository node sends the call result to any slave digital object repository node in the correct slave digital object repository node, for example: and the second slave digital object warehouse node sends a state data request, and the master digital object warehouse node receives the total state data for the multi-point data language running on the second slave digital object warehouse node sent by the second slave digital object warehouse node, so that the memory state data for the multi-point data language running on the master digital object warehouse node is recovered.
For the case that the digital object warehouse node of the number Rayleigh operator is offline and online again due to various factors: under the condition that a first slave digital object warehouse node is offline and online again, the first slave digital object warehouse node sends recovery information to a master digital object warehouse node, and the master digital object warehouse node sends full state data for a multi-point data language running per se to the first slave digital object warehouse node or sends incremental state data for the multi-point data language running per se to the first slave digital object warehouse node or does not send any data to the first slave digital object warehouse node according to the recovery information and the current running mode of the data language cluster; thereby recovering memory state data for the multi-point data language running on the first slave digital object warehouse node.
Under the condition that the master number Rayleigh-end digital object warehouse node is offline and online again, and the master number Rayleigh-end digital object warehouse node is changed into a slave number Rayleigh-end digital object warehouse node, the master number Rayleigh-end digital object warehouse node sends recovery information to the second slave number Rayleigh-end digital object warehouse node, and the second slave number Rayleigh-end digital object warehouse node sends full-quantity state data for a self-running multi-point data language to the master number Rayleigh-end digital object warehouse node or sends incremental state data for the self-running multi-point data language to the master number Rayleigh-end digital object warehouse node or does not send any data to the master number Rayleigh-end digital object warehouse node according to the recovery information and the current running mode of the data language clusters; thereby recovering the memory state data of the multi-point data language running on the main number and the rake operator digital object warehouse node.
Under the condition that the main number rayleigh operator digital object warehouse node is offline and online again and still is the main number rayleigh operator digital object warehouse node, the main number rayleigh operator digital object warehouse node only needs to execute the steps of: and receiving a call request and executing.
Through the multiple recovery methods, the memory state data for the data language can be automatically recovered after the problems of crashing, downtime, power failure and the like of the digital object of the related equipment and the recovery of the related equipment are guaranteed in the operation life cycle of the multi-point data language.
Drawings
Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to designate like parts throughout the figures. In the drawings:
FIG. 1 is a flow chart of a method for recovering memory state data for a data language according to an embodiment of the present invention;
FIG. 2 is a block diagram of a system for recovering memory state data for a data language according to an embodiment of the present invention;
FIG. 3 is a schematic diagram illustrating results obtained by performing fault recovery tests on three recovery methods for data languages with different state quantity sizes under the same network delay in the embodiment of the present invention;
fig. 4 is a schematic diagram illustrating the results obtained by performing fault recovery test on data languages with the same state quantity under different TCP packet delays according to the embodiment of the present invention.
Detailed Description
In order that the above-recited objects, features and advantages of the present invention will become more readily apparent, a more particular description of the invention will be rendered by reference to the appended drawings and appended detailed description. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
The method for recovering memory state data for a data word according to the embodiment of the present invention is provided for a current memory-based calculation mode, referring to fig. 1, a flowchart of a method for recovering memory state data for a data word according to an embodiment of the present invention is shown, where the recovery method is applied to a plurality of digital object warehouse nodes for a plurality of digital operators in a digital network, each of the plurality of digital object warehouse nodes for a plurality of digital operators operates on a data word, the data word is for a multi-point data word, and the plurality of digital object warehouse nodes for a plurality of digital objects form a data word cluster, and the recovery method includes:
step 101: and receiving and executing the call request by the master number Rayleigh operator digital object warehouse node in the data language cluster, and broadcasting the call request to all the slave number Rayleigh operator digital object warehouse nodes in the data language cluster by the master number Rayleigh operator digital object warehouse node.
In the embodiment of the invention, the multi-point data language is generally arranged and operated on a plurality of digital object warehouse nodes of a digital network by the characteristics of the multi-point data language. Because of the scalability and low cost of implementing the implementation of multi-drop data-based data-in-speech (RDS) by using data-in-speech as granularity fragmentation in a memory-based RDS system, the multi-drop data word is thus used at start-up, multiple minority rake digital object repository nodes may be defined by the user themselves to run multi-point data usages.
As one example: when a user needs to use the multi-point data language, 9 multi-point data language uses can be selected by the multi-point data language user, or 20 multi-point data language uses can be selected by the multi-point data language user.
Before the initiation of the multi-point data language, the user needs to select to initiate the request for initiating the multi-point data language on a certain digital object warehouse node, typically using the digital object warehouse node as the main digital object warehouse node in the data language cluster, and the user also needs to select how many digital object warehouse nodes of the multi-point data language are operated on, and the digital object warehouse nodes of the multi-point data language except the main digital object warehouse node of the multi-point data language are slave digital object warehouse nodes in the data language cluster. Along with the above examples: assuming that a user selects to initiate a request for starting and running a multi-point data language on one of the plurality of data object warehouse nodes and selects 10 total plurality of data object warehouse nodes to run the data language, the data object warehouse node initiating the request for starting and running the multi-point data language is taken as a master data object warehouse node, and 9 data object warehouse nodes are arbitrarily selected from the rest 19 data object warehouse nodes as slave data object warehouse nodes.
After the multi-point data language is started and operated, the master number and the plurality of data language digital object warehouse nodes can determine the operation mode of the data language cluster according to the classification of the multi-point data language, wherein the classification of the multi-point data language comprises the following steps: stateless data languages and stateful data languages. The stateless data language refers to: the execution result of the data word is related to the call only, and is not related to the historical state of the data word, for example: four arithmetic data languages, etc.; the stateful data language means: the execution result of the data language is related to its historical execution, for example: counter data words, transfer data words, etc.
If the classification of the multi-point data language is the stateless data language, the main number and Rayleigh operator digital object warehouse node determines that the operation mode of the data language cluster is a common operation mode. If the multi-point data language is classified into the stateful data language, determining that the operation mode of the data language cluster is a common operation mode or a stable operation mode according to the total number of the digital object warehouse nodes of the number Rayleigh operator in the data language cluster and the total number of the digital object warehouse nodes of the number Rayleigh operator in the current normal operation.
Specific: digital object warehouse node total number of number Rate operator in data language cluster between the total number of the digital object warehouse nodes of the number Rayleigh operator which is normally operated currently, under the condition that the first preset condition is met, determining that the operation mode is a common operation mode by the primary number and rake operator digital object warehouse node; digital object warehouse node total number of number Rate operator in data language cluster between the total number of the digital object warehouse nodes of the number Rayleigh operator which is normally operated currently, and under the condition that the second preset condition is met, determining the operation mode to be a stable operation mode by the primary number and data operator digital object warehouse node.
Wherein the first preset condition is: the total number of the digital object warehouse nodes of the number Rayleigh operator in the current normal operation is larger than the total number after the preset operation is carried out on the total number of the digital object warehouse nodes of the number Rayleigh operator in the data language cluster; the second preset condition is: the total number of the digital object warehouse nodes of the number Rayleigh operator in the current normal operation is not more than the total number after the preset operation is carried out on the total number of the digital object warehouse nodes of the number Rayleigh operator in the data language cluster; the preset operation is as follows: and taking the total number of the digital object warehouse nodes of the number Rayleigh operator in the data language cluster as half to obtain a first result, and rounding the first result upwards. Assume that: the total number of the digital object warehouse nodes of the number of the Rayleigh operators in the data language cluster is N, and the total number of the digital object warehouse nodes of the number Rayleigh operators in the data language cluster which are normally operated at present is M; the preset operation is: ⌈ N/2 ⌉, the first preset condition is: m is larger than ⌈ N/2 ⌉, and the second preset condition is that: m is less than or equal to ⌈ N/2 ⌉.
That is, when M is greater than ⌈ N/2 ⌉, it can be determined that the majority of the digital object warehouse nodes of the data language clusters currently run normally, so that the main digital object warehouse nodes of the data language clusters determine that the running mode of the data language clusters is the normal running mode; when M is less than or equal to ⌈ N/2 ⌉, the number of the current normal operation digital object warehouse nodes of the data language cluster can be determined to be less, so that the main digital object warehouse node determines that the operation mode of the data language cluster is a stable operation mode.
After determining the operation mode of the data language cluster, each data Rayleigh operator digital object warehouse node in the data language cluster can store the memory state data of the multi-point data language running by itself according to the operation mode. Different operation modes correspond to different storage methods.
Specific: under the condition that the operation mode is a common operation mode, each digital object warehouse node stores the respective memory state data in a memory persistence mode; under the condition that the operation mode is a stable operation mode, each digital object warehouse node stores the respective memory state data in a hard disk persistence mode.
In the embodiment of the invention, the memory state data for the multi-point data language comprises: full state data and delta state data. The full amount state data includes: all global variables in the data language, current value, data language ID, data language Name, data language type, and data language status version number. The delta state data includes: and calling parameters and method names of the data language request, and if the call contains the cross-data language call, further comprising an execution result of the cross-data language call initiated by the data language.
All global variables and current values in the data language are data language states for the multi-point data language, and in the data language based on memory calculation, the data language states are names, types and values of the global variables in the data language. The change in data-language status is due to the fact that different data-language calls have an effect on the value of the global variable in the data-language. These history states will affect the subsequent data utterance calls.
Status version number for data language: for stateful data usages, each data usages call may cause a change in the state of the data usages. Thus, after each call, a new state version number is generated. For example, the initial state version number is-1, and after the 1 st call request is executed, the state version number becomes 0; after the 2 nd call request is executed, the state version number becomes 1. The state version number can be used for knowing whether the data language state of a certain number of the digital object warehouse nodes is behind other nodes. And when executing a certain data language calling request, judging whether the data Rate digital object warehouse node leaks certain calling requests or repeatedly receives certain calling requests according to whether the current data language version number and the request serial number are continuous.
Cross-data-language calls are calls that support one data-language use to another data-language use in exponential networking, i.e., cross-data-language use calls. The call for one single point data language can be used for one multi-point data language, or the call for another multi-point data language can be used for one multi-point data language.
The meta information for the data language comprises: the data language ID, the data language Name, the data language item Name, the data language state version number, the data language midspan data language calling random variable seed supporting the cross data language calling, and the current cross data language calling random variable value. Whether stateful or stateless, the meta information for the data words needs to be consistent.
The incremental status data is a record of incremental changes in the status of the multi-point data language, as compared to the full status data, and does not reflect the full status information of a multi-point data language at a certain time. An incremental piece of state data can only record the change of the multi-point data words from one state to another.
The specific flow of incremental status data recording is as follows:
(1) The user initiates a call;
(2) Starting incremental state recording at the same time of starting data language execution;
(3) Recording a calling method for the data language and an incoming parameter;
(4) If the method called at this time contains cross-data-language calling, the results of calling other data-language in the cross-data-language calling need to be recorded;
(5) The data word will form a complete incremental status data after completion of the call.
Based on the theory, in the normal operation mode, since the number of the data language clusters which normally operate currently is the majority of the digital object warehouse nodes, while the situation that all the digital object warehouse nodes of the number and the number are crashed at the same time to cause the memory state data to be lost is negligible, therefore, the performance factor of the call for the multi-point data language needs to be prioritized, and all memory state data for the multi-point data language is in the memory in the normal operation mode. In the normal operation mode, each digital object warehouse node of the data Rate does not need to do additional operation, in the execution process of the multi-point data language, the change of the transaction request can cause the change of the memory state data, all the state data of the multi-point data language are in the memory, and the mode is the record of the memory state data of the multi-point data language operated in a memory lasting mode, which is equivalent to the storage of the memory state data of the multi-point data language. Even if the memory persistence has the problem that the memory state data is easy to lose, the correctness of the memory state data for the multi-point data language can be ensured because the number of the current normal operation number of the data language clusters is the majority of the number of the data object warehouse nodes. And after the normal operation is resumed, the correct memory state data can be obtained based on the normal operation digital object warehouse node, so as to ensure the normal operation of the multi-point data language.
In the stable operation mode, because the number of the digital object warehouse nodes of the number of the Rayleigh operators which are normally operated currently in the data language cluster is not more, the availability of data language clusters is prioritized, so each data Rate digital object repository node is stored in a hard disk persistent manner. In general, when the digital object warehouse node of the digital Rayleigh operator stores, the memory state data needs to be recorded based on other programs or modes, so that the hard disk storage of the memory state data is realized. That is, it is necessary to store the memory state data for the data language in the hard disk while executing the multi-point data language. If a data packet loss problem occurs in a data packet storage node, the data packet storage node may recover the memory state data from the hard disk, and the correct memory state data can be acquired based on the normal running digital object warehouse node of the digital data Rayleigh operator, so that the normal running of the multi-point data language is ensured.
In the embodiment of the invention, the specific method for storing each digital object warehouse node by adopting a hard disk persistence mode comprises the following steps: recording full state data or incremental state data for the multi-point data language into a hard disk of the device; when the recording quantity of the increment state data for the multi-point data language reaches a second preset quantity, updating and recording the full quantity state data for the multi-point data language, and deleting the increment state data record for the multi-point data language.
Considering the working efficiency of the data language cluster, the data size of the full-volume state data is generally larger, and the data size of the incremental state data is smaller. And the recording of the full state data in the hard disk also requires the calling of a corresponding method to collect the various variables and their values in the data language memory. Therefore, after the full-quantity state data is recorded to the hard disk storage of the digital object warehouse node of the data Rayleigh operator at a certain moment, when one piece of incremental state data appears later, only the piece of incremental state data is recorded to the hard disk storage, and the working efficiency of the data word cluster is improved. And when the recording quantity of the increment state data reaches a second preset quantity, recording the latest full quantity of state data to a hard disk for storage, and deleting the increment state data record stored in the front. For example: and if the second preset number is 20, recording the full-quantity state data to the hard disk storage at a certain moment by the digital object warehouse node of the number and the number, and then recording the first increment state data to the hard disk storage when the first increment state data appears, recording the second increment state data to the hard disk storage when the second increment state data appears, and the like until the 20 th increment data appears, recording the latest full-quantity state data to the hard disk storage, and simultaneously deleting the 20 increment state data records stored before and the old full-quantity state data record.
It should be noted that, for stateless data language use, since the call result is irrelevant to the history state, if the data language cluster operates in the stable mode, it is meaningless for the stateless data language use to record the full amount of state data plus the increment state data to the hard disk storage, and an additional overhead is caused to the data language cluster. For stateless data language, the data language clusters always operate in normal mode, but not in steady mode.
After all the digital object warehouse nodes of the data language cluster start to operate based on the method, the main digital object warehouse node generally receives the call request sent by the user, and immediately executes the call request after receiving the call request, and meanwhile, the main digital object warehouse node also needs to broadcast the call request to all the slave digital object warehouse nodes of the data language cluster. All slave digital object warehouse nodes in the data language cluster can execute immediately after receiving the call request, and after the execution is finished, the respective call results can be fed back to the master digital object warehouse node.
Step 102: and the master digital object warehouse node receives all calling results fed back from the slave digital object warehouse nodes.
In the embodiment of the invention, the master number and rayleigh digital object warehouse node receives the calling results fed back by all the slave nodes, and naturally, the calling results of the master number and rayleigh digital object warehouse node. The principal number rake operator digital object repository node will compare all call results.
Step 103: and under the condition that the calling result fed back by the first slave digital object warehouse node in all slave digital object warehouse nodes is wrong, the master digital object warehouse node sends the full-quantity state data of the multi-point data language running by the master digital object warehouse node to the first slave digital object warehouse node.
In the embodiment of the invention, when the master digital object warehouse node compares all the calling results, if the calling result fed back by the first slave digital object warehouse node is found to be wrong, the cause of the wrong calling result may be the collapse of the data language used by the multi-point data language running on the first slave digital object warehouse node, or may be the inconsistency between the serial number of the cross-data-language calling and other digital object warehouse nodes, or may be the inconsistency between the current state version number of the multi-point data language and other digital object warehouse nodes. At this time, when the memory state data for the multi-point data language running on the first slave digital object warehouse node is described as being in error, the master digital object warehouse node will send the full-scale state data for the multi-point data language running on the master digital object warehouse node to the first slave digital object warehouse node. Memory state data for the multi-point data language running on the first slave digital object warehouse node is recovered.
Step 104: and under the condition that the master digital object warehouse node executes the calling request and determines that the calling result is wrong, the master digital object warehouse node sends a state data request to a second slave digital object warehouse node in all slave digital object warehouse nodes, wherein the second slave digital object warehouse node is any slave digital object warehouse node in all slave digital object warehouse nodes with correct calling results.
In the embodiment of the invention, when the master number of the multi-point data language is compared with all the calling results, if the calling results are found to be wrong, the reason of the calling result error is the same as that of the first slave number of the multi-point data language, which is possibly caused by the collapse of the data language used by the multi-point data language running on the master number of the multi-point data language, or the inconsistency between the serial number of the cross-data language call and the other number of the multi-point data language, or the inconsistency between the current state version number of the multi-point data language and the other number of the multi-point data language. At this time, it is explained that the memory state data for the multi-point data language running on the master digital rake operator digital object warehouse node is wrong, and then the master digital rake operator digital object warehouse node needs to send a state data request to any slave digital rake operator digital object warehouse node with the correct calling result. For example: and if any slave node with the correct calling result is the second slave node, the master digital object warehouse node sends a state data request to the second slave digital object warehouse node.
Step 105: the master digital data object warehouse node receives the full-quantity state data transmitted by the second slave digital data object warehouse node and used for multi-point data languages running on the second slave digital data object warehouse node.
In the embodiment of the invention, after the second slave digital object warehouse node receives the state data request, the state data of the whole quantity of the multi-point data language used by the second slave digital object warehouse node is sent to the master digital object warehouse node. Memory state data for the multi-point data language running on the primary number rake operator digital object warehouse node is recovered.
Step 106: under the condition that the first slave digital object warehouse node is offline and online again, the first slave digital object warehouse node sends recovery information to the master digital object warehouse node, and the master digital object warehouse node sends full state data for the multi-point data language running by itself to the first slave digital object warehouse node or sends incremental state data for the multi-point data language running by itself to the first slave digital object warehouse node or does not send any data to the first slave digital object warehouse node according to the recovery information and the current running mode of the data language cluster.
In the embodiment of the invention, in the actual running process of the multi-point data language, the master digital object warehouse node or the slave digital object warehouse node may be off-line due to factors such as: a certain number of the digital object warehouse nodes are too long in delay caused by network blockage, and although the digital object warehouse nodes are not actually off-line physically, other digital object warehouse nodes are caused to consider the digital object warehouse nodes to be off-line; or the digital object warehouse node of the digital Rayleigh operator is actually and physically offline due to the factors such as downtime or power failure; or the management process of the digital object warehouse node stops running, but the digital object still runs, and although the digital object warehouse node is not actually and physically disconnected, other digital object warehouse nodes still can be caused to consider the digital object warehouse node to be disconnected.
For any slave digital object warehouse node in the data language cluster, for example, a first slave digital object warehouse node, in the case that the first slave digital object warehouse node is offline and online again, the first slave digital object warehouse node actively sends recovery information to the master digital object warehouse node, and the master digital object warehouse node sends full state data for the self-running multi-point data language to the first slave digital object warehouse node or sends incremental state data for the self-running multi-point data language to the first slave digital object warehouse node or does not send any data to the first slave digital object warehouse node according to the recovery information and the current running mode of the data language cluster.
Specific: the recovery information sent by the first slave digital object warehouse node comprises: the operation mode of the data language cluster before the first slave digital object warehouse node is off line and the version number of the multi-point data language running after the first slave digital object warehouse node is on line again. The first case is: when the data language cluster is in a normal operation mode before the first slave digital object warehouse node is offline, and the data language cluster is in any operation mode currently after the first slave digital object warehouse node is offline again, namely, when the first slave digital object warehouse node is offline, the data language cluster is in the normal operation mode, and after the first slave digital object warehouse node is offline again, the master digital object warehouse node determines that the data language cluster is in any operation mode currently, then the master digital object warehouse node only needs to determine whether the version number of the multi-point data language cluster operated after the first slave digital object warehouse node is offline again is the same as the current state version number of the data language cluster.
If the first slave digital object warehouse node considers that the first slave digital object warehouse node is offline due to the fact that the management process is only crashed or other digital object warehouse nodes are not crashed due to network delay, the recovery information sent to the master digital object warehouse node by the first slave digital object warehouse node carries the version number for the multi-point data language operated after the first slave digital object warehouse node is online again; if the first slave digital object warehouse node breaks down the multi-point data language version number, the recovery information sent from the first slave digital object warehouse node to the master digital object warehouse node does not carry the multi-point data language version number operated after the first slave digital object warehouse node is on line again.
After receiving the recovery information, if the recovery information carries the version number for the multi-point data language operated after the first slave digital object warehouse node is on line again, the master digital object warehouse node can determine whether the version number for the multi-point data language operated after the first slave digital object warehouse node is on line again is the same as the current state version number of the data language cluster; if the recovery information does not carry the version number of the multi-point data language running after the first slave digital object warehouse node is on line again, the master digital object warehouse node directly determines that the version number of the multi-point data language running after the first slave digital object warehouse node is on line again is different from the current state version number of the data language cluster.
If the version number of the multi-point data language used by the first slave digital object warehouse node after being on line again is the same as the current state version number of the data language used group, the master digital object warehouse node does not send any data to the first slave digital object warehouse node. This is because if the current state version number of the multi-point data language for which the first slave digital object repository node is on-line again is the same as the current state version number of the data language for which the first slave digital object repository node is on-line again, then the multi-point data language for which the first slave digital object repository node is on-line again represents that the multi-point data language for which the first slave digital object repository node is on-line does not make any call request, and therefore the first slave digital object repository node does not need to recover the memory state data for the multi-point data language.
If the master digital data object warehouse node determines that the current state version number of the multi-point data language running after the first slave digital data object warehouse node is on line again is different from the current state version number of the data language cluster, the master digital data object warehouse node sends the total state data of the multi-point data language running by the master digital data object warehouse node to the first slave digital data object warehouse node. This is because if the current state version number of the multi-point data language used by the first slave digital object warehouse node after being on line again is different from the current state version number of the data language used group, the multi-point data language used by the data language used group makes at least one call request during the period that the first slave digital object warehouse node is off line and on line again, and because the first slave digital object warehouse node is in a normal operation mode before being off line, the first slave digital object warehouse node has no state data record in a hard disk, the master digital object warehouse node sends the total state data of the multi-point data language used by the master digital object warehouse node to the first slave digital object warehouse node, so that the first slave digital object warehouse node runs the multi-point data language used by the master digital object warehouse node based on the received total state data, and the memory state data of the multi-point data language used by the first slave digital object warehouse node can be recovered.
In the second case: under the condition that the data language cluster is in a stable operation mode before the first slave digital object warehouse node is offline and the data language cluster is in a normal operation mode after the first slave digital object warehouse node is offline again, the master digital object warehouse node is required to send the full state data for the multi-point data language operated by the master digital object warehouse node to the first slave digital object warehouse node, so that the first slave digital object warehouse node can operate the multi-point data language of the master digital object warehouse node based on the received full state data.
Third case: under the condition that the cluster for the data words before the first slave digital object warehouse node is offline is in a stable operation mode and the cluster for the data words after the first slave digital object warehouse node is offline is in the stable operation mode, the first slave digital object warehouse node acquires full state data and incremental state data for the multi-point data words from a hard disk of the first slave digital object warehouse node, meanwhile, the master digital object warehouse node determines whether the version number for the multi-point data words operated after the first slave digital object warehouse node is offline is the same as the current state version number of the cluster for the data words, if the version number for the multi-point data words is the same, the master digital object warehouse node does not send any data to the first slave digital object warehouse node, if the version number for the multi-point data words operated by the master digital object warehouse node is different, the master digital object warehouse node sends the new incremental state data for the multi-point data words operated by the master digital object warehouse node to the first slave digital object warehouse node, so that the first slave digital object warehouse node is based on the received new incremental state data and the full state data and the acquired full state data and the incremental state data for the multi-point data operated by the master digital object warehouse node, wherein the master digital object warehouse node is the new incremental state data for the multi-point data operated by the slave digital object warehouse node: the first slave digital object warehouse node has no incremental state data in its own hard disk caused by offline.
In the third case, since the cluster for data words before the first slave digital object warehouse node goes offline is in the stable operation mode, the first slave digital object warehouse node stores the full state data and the incremental state data for the multi-point data words running by itself into its own hard disk, and when the first slave digital object warehouse node resumes online, the first slave digital object warehouse node can restore the full state data and the incremental state data from its own hard disk and then send restoration information to the master digital object warehouse node because the full state data and the incremental state data before going offline are stored in the hard disk. After the master digital object warehouse node receives the recovery information, the same method as in the first case is adopted to determine whether the state version numbers are the same, if the state version numbers are the same, the data language clusters have no incremental state data in the period of offline and online again of the first slave digital object warehouse node, and if the state version numbers are not the same, the data language clusters have new incremental state data in the period of offline and online again of the first slave digital object warehouse node, so that the master digital object warehouse node needs to send the new incremental state data which is not stored in the hard disk of the first slave digital object warehouse node to the first slave digital object warehouse node.
Regarding how the operation mode of the data language cluster is dynamically switched when the first slave digital object warehouse node is offline and online again, or when the master digital object warehouse node is offline and online again, the description is omitted below.
Step 107: and under the condition that the master number Rayleigh-end digital object warehouse node is offline and online again and the master number Rayleigh-end digital object warehouse node is changed into a slave number Rayleigh-end digital object warehouse node, the master number Rayleigh-end digital object warehouse node sends recovery information to the second slave number Rayleigh-end digital object warehouse node, and the second slave number Rayleigh-end digital object warehouse node sends full-quantity state data for the self-running multi-point data words to the master number Rayleigh-end digital object warehouse node or sends incremental state data for the self-running multi-point data words to the master number Rayleigh-end digital object warehouse node or does not send any data to the master number Rayleigh-end digital object warehouse node according to the recovery information and the current running mode of the data words.
In the embodiment of the present invention, the foregoing step 106 illustrates how to recover the memory state data for the multi-point data language after the first slave digital object warehouse node serving as any slave digital object warehouse node is offline and online again. And after the master number rake digital object warehouse node is offline and online again, the second slave node becomes a new master number rake digital object warehouse node, and the master number rake digital object warehouse node becomes a slave number rake digital object warehouse node how to recover the memory state data used by the multi-point data language, which is basically the same as the first slave node.
Under the condition that the data language cluster before the main number and the sub number of the sub number and the sub number are off line is in a common operation mode and the data language cluster after the main number and the sub number are on line again is in any operation mode, the second sub number and the sub number of the sub number and the sub number are determined whether the version number of the multi-point data language cluster operated after the main number and the sub number of the sub number and the sub number are the same, and if the data are the same, the second slave digital object warehouse node does not send any data to the master digital object warehouse node, and if the data are not the same, the second slave digital object warehouse node sends the full-quantity state data of the multi-point data language used by the slave digital object warehouse node to the master digital object warehouse node, so that the master digital object warehouse node runs the multi-point data language used by the slave digital object warehouse node based on the received full-quantity state data.
Under the condition that the data language cluster is in a stable operation mode before the main number and the sub-operator digital object warehouse node is off line and the data language cluster is in a common operation mode after the main number and the sub-operator digital object warehouse node is on line again, the second slave digital data object warehouse node sends the total state data of the self-running multi-point data language to the master digital data object warehouse node so that the master digital data object warehouse node runs the self-multi-point data language based on the received total state data.
Under the condition that the data word cluster before the main number of the sub-digital object warehouse node is offline is in a stable operation mode and the data word cluster after the main number of the sub-digital object warehouse node is online again is in the stable operation mode, the main number of the sub-digital object warehouse node acquires full state data and increment state data for the multi-point data word from a hard disk of the main number of the sub-digital object warehouse node, meanwhile, the second slave number of the sub-digital object warehouse node determines whether the version number for the multi-point data word operated after the main number of the sub-digital object warehouse node is online again is the same as the current state version number of the data word cluster, if the version number for the multi-point data word is the same as the current state version number of the data word cluster, the second slave number of the sub-digital object warehouse node does not send any data to the main number of the sub-digital object warehouse node, and if the version number of the sub-digital object warehouse node is not the same, the second slave number of the sub-digital object warehouse node sends the new increment state data for the multi-point data word operated by the main number of the sub-digital object warehouse node to the main number of the sub-digital object warehouse node, so that the main number of the sub-digital object warehouse node can operate the multi-point data word based on the received new increment state data and the acquired full state data and increment state data for the main number of the sub-digital object warehouse node.
During the operation of the actual data language cluster, the primary number and the secondary number can be disconnected and disconnected again. It can be divided into two categories: the first class is the condition that the primary number of the digital object warehouse nodes are offline and online again, and the primary number of the digital object warehouse nodes become slave number of the digital object warehouse nodes and become any slave number of the digital object warehouse nodes, namely the second slave number of the digital object warehouse nodes of the slave number of the digital object warehouse nodes become new primary number of the digital object warehouse nodes; the class 2 is the case where the primary number rayleigh digital object warehouse node is offline and online again, but the primary number rayleigh digital object warehouse node is still the primary number rayleigh digital object warehouse node.
For the first type of case there are the following steps:
step S1: and under the condition that the main number and Rayleigh operator digital object warehouse nodes are offline and the offline time exceeds the preset time and are not online again, other number and Rayleigh operator digital object warehouse nodes in the data language cluster are disconnected from the main number and Rayleigh operator digital object warehouse nodes, and a re-election request is sent to a number and Rayleigh network identification analysis system in the number network.
In the embodiment of the invention, the master digital object warehouse node and the slave digital object warehouse node are kept connected with each other based on a heartbeat mechanism, and whether the other party is off-line or on-line is determined based on the heartbeat mechanism. When the primary number of the digital object warehouse nodes are offline and the offline time period exceeds a preset time period, for example: the preset duration is 5 seconds, then when the master number rake digital object warehouse node is down and the down duration exceeds 5 seconds and has not been re-on, other digital object warehouse nodes of the data language cluster, that is, all slave digital object warehouse nodes determine that the master digital object warehouse node is offline, and all slave digital object warehouse nodes are automatically disconnected from the master digital object warehouse node. And all slave digital object warehouse nodes of the number and Rayleigh operator can send a re-election request to a number and Rayleigh network identification analysis system in the number and Rayleigh network. The digital Rayleigh network identification analysis system is a management node in the digital network, does not participate in operations such as consensus, storage and the like of the digital network, and is only responsible for managing the admission of the digital Rayleigh sub digital object warehouse node and the digital Rayleigh storage digital object warehouse node in the digital network.
Step S2: after receiving the re-election request sent by all the digital object warehouse nodes except the main digital object warehouse node, the digital networking identification analysis system elects a second slave digital object warehouse node from all the digital object warehouse nodes except the main digital object warehouse node as a new main digital object warehouse node;
step S3: after the second slave digital data object warehouse node becomes the new master digital data object warehouse node, a connection request is sent to all digital data object warehouse nodes except the master digital data object warehouse node, so that all digital data object warehouse nodes except the master digital data object warehouse node are connected with the second slave digital data object warehouse node.
In the embodiment of the invention, after receiving the re-election request sent by all the digital object warehouse nodes except the main digital object warehouse node, the digital networking identification analysis system takes the second slave digital object warehouse node as a new main digital object warehouse node from all the digital object warehouse nodes except the main digital object warehouse node. Even during this period, the primary number of rayleigh digital object warehouse node has been re-brought online, but because of its occurrence of the process of coming offline and re-bringing online, the number of rayleigh digital network identification resolution system considers that the primary number of rayleigh digital object warehouse node has a stability problem, based on the stability consideration of the number of network operations, the primary number of rayleigh digital object warehouse node as the primary number of rayleigh digital object warehouse node will not be based on the computer.
After the second slave digital end object warehouse node becomes the new master digital end object warehouse node, a connection request is sent to all digital end object warehouse nodes except the master digital end object warehouse node so that all digital end object warehouse nodes except the master digital end object warehouse node are connected with the second slave digital end object warehouse node.
Step S4: and after the master number regex digital object warehouse node is on line again, the second slave number regex digital object warehouse node sends a connection request to the master number regex digital object warehouse node so that the master number regex digital object warehouse node connects the second slave number regex digital object warehouse node with the identity of the slave number regex digital object warehouse node.
After the second slave digital object warehouse node becomes the new master digital object warehouse node, after the master digital object warehouse node is on line again, the master digital object warehouse node queries that the current master digital object warehouse node is the second slave digital object warehouse node, and then the master digital object warehouse node is actively connected with the second slave digital object warehouse node. Such that the master digital data object repository node stores slave digital data objects the identity of the warehouse node is connected with the second slave digital object warehouse node. Of course, if the master number rayleigh digital object warehouse node has been newly brought online when the second remade node becomes the new master number rayleigh digital object warehouse node, then after the second slave number rayleigh digital object warehouse node becomes the new master number rayleigh digital object warehouse node, a connection request is sent to all the number rayleigh digital object warehouse nodes except the master number rayleigh digital object warehouse node, and a connection request is also sent to the master number rayleigh digital object warehouse node.
Step 108: and under the condition that the main number rayleigh digital object warehouse node is offline and online again and still is the main number rayleigh digital object warehouse node, executing by the main number rayleigh digital object warehouse node: and receiving and executing the call request, and broadcasting the call request to all slave digital object warehouse nodes in the data language cluster by the master digital object warehouse node.
In the embodiment of the invention, for the second class of cases: under the condition that the main number of the digital object warehouse nodes are offline and the offline time period does not exceed the preset time period, other digital object warehouse nodes of the data language cluster cannot sense the offline of the main number of the digital object warehouse nodes, connection with the main number of the digital object warehouse nodes is still maintained, and the main number of the digital object warehouse nodes is still the main number of the digital object warehouse nodes.
And after the primary number and the end operator digital object warehouse node are online again, executing the steps of: and receiving and executing the call request, and broadcasting the call request to all slave digital object warehouse nodes in the data language cluster by the master digital object warehouse node.
In the embodiment of the invention, if the primary number and the secondary number are taken off line from the digital object warehouse node, and the off-line time does not exceed the preset time, the digital object warehouse node is taken on line again. Because the offline time is extremely short, other digital object warehouse nodes of the data language cluster cannot sense that the main digital data object warehouse node is offline, still keeps connection with the main digital data object warehouse node, and cannot send a re-election request like a digital data networking identification analysis system. Therefore, after the primary number and end operator digital object warehouse node is online again, no matter whether the operating multi-point data language memory state data is normal or not, the primary number and end operator digital object warehouse node only needs to execute step 101: and receiving and executing the call request, and broadcasting the call request to all slave digital object warehouse nodes in the data language cluster by the master digital object warehouse node. After the primary digital object warehouse node is on line again, the next call request needs to be waited, and the memory state data for the multi-point data language can be recovered by continuing to step 102-step 105.
In the embodiment of the invention, for the running state of the data language cluster, when the data language cluster starts to run, the main digital rake operator digital object warehouse node can determine the current running mode of the data language cluster according to the condition that the total number of the digital rake operator digital object warehouse nodes in the data language cluster is between the total number of the digital rake operator digital object warehouse nodes and the total number of the digital rake operator digital object warehouse nodes which normally run at present, and the first preset condition or the second preset condition is met. In the running process of the data language cluster, any one or more slave digital object warehouse nodes or master digital object warehouse nodes are problematic due to factors in various aspects, so that the relation between the total number of the digital object warehouse nodes of the data language cluster and the total number of the digital object warehouse nodes of the current normal operation is changed, or after the data language cluster is restored again, the relation between the total number of the digital object warehouse nodes of the data language cluster and the total number of the digital object warehouse nodes of the current normal operation is changed, and the master digital object warehouse nodes can determine whether the data language cluster needs to be switched in running mode. Specific:
When the data language cluster starts to operate, a first preset condition is met between the total number of the digital object warehouse nodes of the number Rayleigh operator and the total number of the digital object warehouse nodes of the number Rayleigh operator in the data language cluster, the main digital object warehouse node determines that the current operation mode of the data language cluster is a common operation mode, and the main digital object warehouse node and all the slave digital object warehouse nodes of the number Rayleigh operator adopt a memory persistence mode to store memory state data. Assuming that the first slave digital object warehouse node drops off line, then it is equivalent to that the data language cluster operation mode is a normal operation mode before the first slave digital object warehouse node drops off line, and the first slave digital object warehouse node stores the memory state data in a memory persistence mode.
And when the first slave digital object warehouse node is offline and is not online again, the current cluster data language still meets the condition of being in the normal operation mode, and then the master digital object warehouse node keeps the operation mode still to be in the normal operation mode. In another case, when the first slave digital object warehouse node is offline and has not been online again, the current cluster data language does not meet the condition of being in the normal operation mode, and then the master digital object warehouse node switches the operation mode to the stable operation mode.
When the first slave digital object warehouse node is offline and is not online again, and the operation mode is the stable operation mode, after the first slave digital object warehouse node is online again, if the total number of the digital object warehouse nodes of the data language cluster is between the total number of the digital object warehouse nodes of the data language cluster and the total number of the digital object warehouse nodes of the current normal operation, the main digital object warehouse node switches the operation mode into the normal operation mode. Naturally, if the condition in the normal operation mode is still not satisfied between the total number of the digital object warehouse nodes of the number Rayleigh operator and the total number of the digital object warehouse nodes of the number Rayleigh operator in the data language cluster after the first slave digital object warehouse node is on line again, the operation mode of the master digital object warehouse node is still the stable operation mode.
In addition, it should be noted that, only after the primary number and the reserve operator digital object warehouse node switches the operation mode from the stable operation mode to the normal operation mode, the normal operation mode is the memory persistence mode, the hard disk persistence mode record is not needed, in order to save the hard disk storage space of the digital object warehouse nodes of the data language cluster, all the digital object warehouse nodes of the data language cluster need to delete the full state data and the incremental state data recorded in the hard disk during the stable operation mode after the operation mode is switched to the normal operation mode.
In summary, through the above-mentioned multiple recovery methods, it is ensured that after the problems of breakdown, downtime, power failure, etc. of the digital object of the computer node occur and normal recovery occurs in the operation life cycle of the multi-point data language, the memory state data of the data language can be automatically recovered.
Based on the above method for recovering the memory state data for the data language, the embodiment of the invention also provides a system for recovering the memory state data for the data language, which comprises: a plurality of digital object repository nodes of a digital network, the plurality of digital object repository nodes of the digital network comprising: a master number of rayleigh digital object warehouse nodes and a slave number of rayleigh digital object warehouse nodes; the data language is used on each of the plurality of data-to-data-operator digital object warehouse nodes, the data language is used as a multi-point data language, the plurality of data-to-data-operator digital object warehouse nodes form a data language cluster, referring to fig. 2, a block diagram of a main data-to-data-operator digital object warehouse node in a recovery system of memory state data for data languages according to an embodiment of the present invention is shown, and the main data-to-data-operator digital object warehouse node includes:
The receiving and executing broadcasting module is used for receiving and executing the calling request and broadcasting the calling request to all slave digital object warehouse nodes in the data language cluster;
the receiving result module is used for receiving all calling results fed back from the digital object warehouse node of the digital Rayleigh operator;
the first sending module is used for sending the full-quantity state data of the multi-point data language running by the first slave digital object warehouse node to the first slave digital object warehouse node under the condition that the calling result fed back by the first slave digital object warehouse node in all slave digital object warehouse nodes is wrong;
the data request module is used for sending a state data request to a second slave digital object warehouse node in all slave digital object warehouse nodes under the condition that the calling request is executed and the calling result of the data request module is determined to be wrong, wherein the second slave digital object warehouse node is any slave digital object warehouse node in all slave digital object warehouse nodes with correct calling results;
a data receiving module, configured to receive full-quantity state data for a multi-point data language running on the second slave digital object warehouse node, where the full-quantity state data is sent by the second slave digital object warehouse node;
The receiving information sending data module is used for receiving recovery information sent by the first slave digital object warehouse node when the first slave digital object warehouse node is offline and online again, sending full state data for the self-operated multi-point data language to the first slave digital object warehouse node or sending the incremental state data for the self-operated multi-point data language to the first slave digital object warehouse node or not sending any data to the first slave digital object warehouse node according to the recovery information and the current operation mode of the data language cluster;
the information sending and receiving data module is used for sending the recovery information to the second slave digital object warehouse node under the condition that the second slave digital object warehouse node is offline and online again and becomes a slave digital object warehouse node, and the second slave digital object warehouse node becomes a new master digital object warehouse node, and receiving the total state data or the incremental state data for the multi-point data words operated by the second slave digital object warehouse node according to the recovery information and the operation mode;
The execution module is used for executing the steps under the condition that the execution module is offline and online again and is still a master number and Rate operator digital object warehouse node: and receiving and executing the call request, and broadcasting the call request to all slave digital object warehouse nodes in the data language cluster.
Optionally, the primary number rake digital object warehouse node further includes: a determine operation mode module for:
under the condition that the classification of the multi-point data language is a stateless data language, determining that the operation mode is a common operation mode;
determining that the operation mode is the common operation mode when the classification of the multi-point data language is a stateful data language and the total number of the digital object warehouse nodes of the number Rayleigh operator in the data language cluster is between the total number of the digital object warehouse nodes of the number Rayleigh operator in the current normal operation and the total number of the digital object warehouse nodes of the number Rayleigh operator in the current normal operation meets a first preset condition;
determining that the operation mode is the stable operation mode under the condition that the classification of the multi-point data language is used as a stateful data language, and the second preset condition is met between the total number of the digital object warehouse nodes of the number of the Rayleigh operators in the data language cluster and the total number of the digital object warehouse nodes of the number of the Rayleigh operators in the current normal operation;
The master digital data object warehouse node and the slave digital data object warehouse node each comprise: the operation record module is used for:
under the condition of the common operation mode, operating in a memory persistence mode;
under the condition of the stable operation mode, recording the full-quantity state data and the increment state data for the multi-point data language into a hard disk of the device;
when the recording quantity of the incremental state data for the multi-point data language reaches a preset quantity, respectively updating and recording the full state data for the multi-point data language, and respectively deleting the incremental state data records for the multi-point data language;
wherein, the first preset condition is: the total number of the current normal running digital object warehouse nodes of the digital speech clusters is larger than the total number of the preset operation on the total number of the digital object warehouse nodes of the digital speech clusters, and the preset operation is as follows: half the total number of the digital object warehouse nodes of the number-of-Rayleigh operator in the data language cluster to obtain a first result, and rounding the first result upwards;
the second preset condition is: the total number of the digital object warehouse nodes of the number Rayleigh operator which is currently and normally operated, and the total number is not larger than the total number after the preset operation is carried out on the total number of the digital object warehouse nodes of the number Rayleigh operator in the data language cluster.
Optionally, the recovery information includes: the operation mode of the data language cluster before the first slave digital object warehouse node is offline and the version number of the multi-point data language operated after the first slave digital object warehouse node is offline again;
the information receiving and data transmitting module is specifically configured to:
when the data language cluster is in the normal operation mode before the first slave digital object warehouse node is offline, and the data language cluster is in any operation mode after the first slave digital object warehouse node is online again, determining whether the version number of the multi-point data language cluster operated after the first slave digital object warehouse node is online again is the same as the current state version number of the data language cluster, if so, not sending any data to the first slave digital object warehouse node, otherwise, sending the full state data of the multi-point data language operated by the first slave digital object warehouse node to the first slave digital object warehouse node, so that the first slave digital object warehouse node operates the multi-point data language of the first slave digital object warehouse node based on the received full state data;
When the data language cluster is in the stable operation mode before the first slave digital object warehouse node is offline and the data language cluster is in the normal operation mode after the first slave digital object warehouse node is online again, transmitting full-quantity state data of the multi-point data language running by the first slave digital object warehouse node to the first slave digital object warehouse node so that the first slave digital object warehouse node runs the multi-point data language by the first slave digital object warehouse node based on the received full-quantity state data;
and under the condition that the data word cluster is in the stable operation mode before the first slave digital object warehouse node is offline, and the data word cluster is in the stable operation mode after the first slave digital object warehouse node is offline again, the first slave digital object warehouse node acquires full-quantity state data and incremental state data for the multi-point data word from a hard disk of the first slave digital object warehouse node, simultaneously determines whether the version number for the multi-point data word running after the first slave digital object warehouse node is offline again is the same as the current state version number of the data word cluster, if the version number for the multi-point data word is the same as the current state version number of the data word cluster, any data is not transmitted to the first slave digital object warehouse node, and if the version number for the multi-point data word running by the first slave digital object warehouse node is different, the first slave digital object warehouse node is enabled to be based on the received new incremental state data and the full-quantity state data and the state data acquired from the hard disk of the first slave digital object warehouse node, and the first slave digital object warehouse node is not enabled to be the incremental state data of the multi-point data word from the hard disk of the first slave digital object warehouse node.
Optionally, the recovery information further includes: the operation mode of the data language cluster before the main number and the plurality of data language version numbers are used for the operation after the main number and the plurality of data object nodes are used for the operation;
the information sending and receiving data module is specifically configured to:
transmitting said recovery information to said second slave digital object warehouse node;
receiving the full state data of the multi-point data language transmitted by the second slave digital object warehouse node of the number and operating the multi-point data language of the slave digital object warehouse node based on the received full state data;
or acquiring full state data and incremental state data for the multi-point data language from the own hard disk, receiving the newly-added incremental state data for the multi-point data language sent by the second slave digital object warehouse node, and running the own multi-point data language based on the received newly-added incremental state data and the full state data and the incremental state data acquired from the own hard disk.
Based on the method for recovering the memory state data for the data language, the embodiment of the invention also provides a digital networking system, which comprises the following steps: the system comprises a Rayleigh network identification analysis system, a plurality of Rayleigh storage digital object warehouse nodes and a plurality of Rayleigh operator digital object warehouse nodes; the digital Rayleigh network identification analysis system is used for managing a plurality of digital object warehouse nodes of the digital Rayleigh operators and a plurality of digital object warehouse nodes of the digital Rayleigh storage operators; for example: annotating any server or user equipment to enter a digital networking system as a digital object warehouse node of a digital Rayleigh operator; the plurality of digital object warehouse nodes are used for executing the recovery method of the data language memory state data in any one of the steps 101-108; the at least one digital object repository node is configured to store operation results for the multi-point data language sent by the plurality of digital object repository nodes.
In order to verify the practicability of the recovery method of the invention, experimental tests and evaluations are performed on the recovery method for the multi-point data language.
1 device is used as a test server, 1 device is used as a digital data networking identification analysis system, 7 devices are used as digital data object warehouse nodes, the North Dacron big data digital networking system is simulated, and is a typical big data digital networking system for operating data languages based on a memory. The application is exemplified as a multi-point data language with a simple key value, which is a data language for creating users and transferring money between users, and any multi-point data language known at present can be used for experiments.
For the test of fault recovery availability of the multi-point data language, the whole thought is as follows:
and testing whether a fault recovery mechanism of the digital object warehouse node of the digital Rayleigh operator can normally operate in a breakdown sequence mode. N is used for representing the number of all the number of the number-of-Rayleigh operator digital object warehouse nodes in the data language cluster, M is used for representing the number of the number-of-Rayleigh operator digital object warehouse nodes in the data language cluster which normally operate, Q is used for representing the number of results which are returned to the main number-of-Rayleigh operator digital object warehouse nodes and are consistent in a certain number of data language calls, and the result is considered to be a correct call result when the result is considered. For each of the crash sequences, a data language call request is initiated to any one of the online data Rate digital object repository nodes in the data language cluster at any time, and the execution result is recorded. After the whole 'crash sequence' execution is completed, judging whether the node is successfully recovered and available by monitoring the recorded execution result of each data language call request.
Taking the example of the collapse sequence "1234567- >123- >12345- >234- >2345" for a cluster of 7 number of rayleigh digital object warehouse nodes, where "1234567" means that 7 number of rayleigh digital object warehouse nodes are all online and "123" means that only 3 number of rayleigh digital object warehouse nodes "123" are online, at this time the number of rayleigh digital object warehouse nodes "4567" are taken out of operation, and then the number of rayleigh digital object warehouse nodes "45" are restarted from "123" to "12345".
And performing test environment construction, program deployment and the like according to experimental requirements, and then performing usability tests on stateful data languages and stateless data languages used in the 5-number Rayleigh operator digital object warehouse node cluster and the 7-number Rayleigh operator digital object warehouse node cluster respectively.
And a collapse sequence file is also required to be generated, and the collapse sequence is written and generated and written by taking the collapse sequence of the 7-number-of-Rayleigh-operator digital object warehouse node data language cluster as an example. When each new sequence is generated, randomly selecting a plurality of currently online digital object warehouse nodes for stopping working, or randomly selecting a plurality of currently crashed digital object warehouse nodes for recovering working. Setting the sequence length to 700 eventually generates a crash sequence of length 700, and the sequence in which the data-language call request can be executed is 665. The collapse sequence covers paths of different recovery methods after all the digital object warehouse nodes of the digital object are on line again. Wherein the partial sequences are shown in the following table:
And the call request file is also required to be generated, and different requests are generated for different data languages and written. Taking a simple key value pair as an example, when a request is generated, methods for creating a user, setting the balance of an initial account, transferring accounts, inquiring the balance of a certain account, inquiring the balance of all accounts, inquiring the total balance and the like are randomly selected, calling parameters are generated according to the selected methods, and the names and the parameters of the methods are separated by division numbers, so that the request is generated.
Usability test for stateful data languages:
two stateful data languages will be tested separately in this section: one is for counter counting data words, which comprises two methods, one is used for incremental calculation, the other is used for accumulating the length of the input parameters, and the state quantity of the data words is smaller; the other is simple key value and the state quantity of the data words is increased along with the increase of the number of accounts, and is larger than that of counter data words.
1) Test for counter count data words
And testing in the 5-number Rayleigh operator digital object warehouse node cluster and the 7-number Rayleigh operator digital object warehouse node cluster respectively, and obtaining a data language calling request result recorded by a testing program.
The counter count data words of a single point are started in other local notebook computers, the execution is carried out through the same request file, and the single point execution result is recorded, so that the correct, wrong and unavailable results in 665 data word call requests executed in the 5-number-of-Rayleigh-operator digital object warehouse node cluster and the 7-number-of-Rayleigh-operator digital object warehouse node cluster are compared by taking the counter count data words as a standard.
And (3) obtaining the situation that the counter counts the data words and is used in the execution results of the data words in the 5-number Rayleigh operator digital object warehouse node cluster and the 7-number Rayleigh operator digital object warehouse node cluster by programming the comparison result, wherein the situation is shown in the following table:
2) Simple key value application test
And testing the application of the simple key value in the same way, and obtaining the call request result for the data language recorded by the test program.
And (3) obtaining the situation of simple key value application to the data language execution results in the 5-number Rayleigh operator digital object warehouse node cluster and the 7-number Rayleigh operator digital object warehouse node cluster by programming the comparison result, wherein the situation is shown in the following table:
3) Usability test for stateless data language
In this section, the mathOperation for data words is used for a stateless data word in charge of addition, subtraction, multiplication and division, and the execution result is only related to the input parameters.
And testing in the 5-number Rayleigh operator digital object warehouse node cluster and the 7-number Rayleigh operator digital object warehouse node cluster respectively, and obtaining a data language calling request result recorded by a testing program.
And (3) obtaining the situation of the execution result of the data language in the 5-number Rayleigh operator digital object warehouse node cluster and the 7-number Rayleigh operator digital object warehouse node cluster by writing the program of the comparison result, wherein the situation is shown in the following table:
the above is a test for usability for a multi-point data language, and the following is a test for failure recovery time for a multi-point data language.
The fault recovery time test for multi-point data languages includes a number of situations: the data languages with different state quantity sizes under the same network delay pass through recovery (recovery from common) based on a common operation mode, recovery (recovery from stable) based on a stable operation mode and recovery (recoverFromStable (need trans from master)) requiring a master number rake operator digital object warehouse node to send newly added incremental state data, and the fault recovery time of the three modes; the data words with the same state quantity are used for fault recovery time in three modes under different network delays; adopting new main number and rayleigh operator digital object warehouse node election time after the main number and rayleigh operator digital object warehouse node crashes under the same network delay when different heartbeat packet periods are adopted; and under the same heartbeat packet period, the new main number and rayleigh operator digital object warehouse node election time is obtained after the main number and rayleigh operator digital object warehouse node crashes under different network delays.
The same fault is repeated 10 times in the test program when the fault recovery time of the digital object warehouse node of the digital Rate is tested, the recovery time of 10 times is recorded, and the final fault recovery time is represented by the average value of the 10 fault recovery times. The application is tested for data language modeling different state quantity sizes using simple key values with different account numbers. Firstly, a single-point simple key value pair application is started through a local notebook computer, a request is sent to the local simple key value pair application through an automatic execution request and a program for saving the state, and the state file of the data language under 2000 users, 4000 users and 6000 users … … and 50000 users is automatically saved. And transmits the state files one by one to the trusted digital rake digital object repository node of the server.
The state file is automatically loaded after the multi-point data language is set to start. Thus, the multi-point data language with different state quantity can be simulated. Similar to the test cluster availability described above, a crash sequence of length 10 is also generated when testing for failure recovery time. And sending a data language call request to each crash sequence to judge whether the normal recovery can be realized, wherein the data language call request is a transfer and balance inquiry request.
In the data language fault recovery mechanism based on memory calculation, two main fault recovery methods exist after the node is on line again: firstly, the full state data used by the current multipoint data language of the digital object warehouse node dump of other digital operators is transmitted to a fault node for recovery; and secondly, recovering the full state data and the incremental state data for the multi-point data language recorded in the local disk. If a digital data object repository node recovers through the full amount of state data and incremental state data for a multi-point data word recorded in a local disk, if the node remains behind after recovering the data through the local disk, for example, during a crash, the data word cluster performs a call request, then the digital data object repository node needs to retrieve from the master digital data object repository node after recovering through the local state data its record of incremental state data missed during the crash.
Under the same network delay, the three recovery methods are subjected to fault recovery test by using the data languages with different state quantity sizes through the method, so that the result shown in the figure 3 is obtained, the abscissa in the figure 3 is the number of users for the data languages, and the ordinate is the time for recovering the data, and the unit is millisecond (ms). The darkest color in fig. 3 is the recover from common method, the darker color is the recover from stable method, and the lightest color is recoverFromStable (need trans from master)) method. It can be known that the recovery from stable method has the least data use time and recoverFromStable (need trans from master)) the recovery from common method has the longest data use time under the same network delay and the same number of users for the data.
And testing the fault recovery time of the data language when the network delays are different.
In this section, the influence of network delay on the fault recovery time of the data language is tested, and the simple key value application with 10000 accounts is adopted as a test object. The network delay is modeled by programming a script to set the TCP packet delay times for all trusted digital rake digital object repository nodes.
By the method, the data languages with the same state quantity under different TCP packet delays are subjected to fault recovery test respectively to obtain the result shown in figure 4. The curve formed by the diamond plus the solid line in fig. 4 is the recover from common method, the curve formed by the triangle plus the solid line is the recover from stable method, and the curve formed by the square plus the solid line is recoverFromStable (need trans from master)). It can be known that, under the condition of data words with the same state quantity size under different TCP packet delays, the recovery from stable method has the least data use time, recoverFromStable (need trans from master)) the method has the longer data use time, and the recovery from common method has the longest data use time.
And (3) testing the election time of the nodes of the primary number and data object warehouse when the heartbeat packet periods are different.
Because the different heartbeat packet periods can cause the difference in time when the digital object warehouse node of the number rayleigh operator discovers that the digital object warehouse node of the main number rayleigh operator crashes, the time influence of different heartbeat packet periods on the time when the digital object warehouse node of the main number rayleigh operator crashes, the new digital object warehouse node of the main number rayleigh operator and the new digital object warehouse node of the main number rayleigh operator finish cluster information data recovery is tested in the section. The period of the heartbeat packet is changed by setting the heartbeat packet and the fault detection period. After setting a new heartbeat packet period, the steps of deploying the trusted digital rake digital object warehouse node data language engine are repeated. And then running a corresponding test program to perform a test, and obtaining the results shown in the following table:
and (3) selecting time test for the nodes of the primary number and the secondary number of the digital object warehouse when the network delays are different.
Because the slave digital object warehouse node monitors that the crash is affected by the network environment after the master digital object warehouse node crashes, and the new master digital object warehouse node election process and the new master digital object warehouse node cluster data recovery all need to interact with network communication for many times, the node will test the network delay for the new master digital object warehouse node recovery time. Setting the heartbeat packet period to 2000ms, adopting the above-mentioned middle mode to simulate different network environments, and testing to obtain the following table results:
In summary, in the same network environment, the failure recovery time for data languages with different state quantity sizes increases as the state quantity size increases. Wherein the recovery from common method takes the longest time and the largest fluctuation with increasing state quantity. This is because the recover from common method requires that the prime number rake digital object repository node dump current data language be used with the full amount of state data and sent. When the state quantity is increased, the time for storing the current full-quantity state data for the data language is gradually increased, and the time for transmitting the full-quantity state data file is also increased due to the increase of the full-quantity state data. The recover from stable method performs state recovery from the local disk, and the performance is superior to that of the recover from common method. Compared with the recover FromStable method, recoverFromStable (need trans from master) needs to additionally acquire the incremental state data which is not recorded during the crash from the primary number rake digital object warehouse node, so that the performance of the method is slightly inferior to that of the recover FromStable method, but is superior to that of the recover FromCommon method.
For the data language with the same state quantity, the recovery from stable method is still the best in recovery performance under different network delay environments. Because the time to transmit the full amount of state data increases with network delay, the time-to-use rise of the recovery from command is most pronounced.
In the same network environment, different heartbeat packet periods can influence the time when the slave digital object warehouse node detects that the master digital object warehouse node breaks down, and further influence the time when a new master digital object warehouse node elects and recovers after the master digital object warehouse node breaks down. As the period of the heartbeat packet increases, the time spent electing new prime number rake digital object warehouse nodes increases gradually.
The time for which the new master digital object repository node recovers is different for data usages of the same heartbeat packet period under different network delays. Network delays can affect the transmission of heartbeat packets and new primary number of rayleigh digital object repository node elections and recovery can involve interactions between multiple number of rayleigh digital object repository nodes. As network delays increase, new dominant digital object warehouse nodes gradually increase in recovery time.
For performance testing:
in this section, the execution performance of the single-point data language and the multi-point data language will be compared by the request execution time for the data language. Taking the simple key value corresponding application with 30000 accounts at the beginning as an example for testing, sending a request for inquiring all accounts and balances and a request for inquiring a balance of a certain account to a single-point data language and a multi-point data language respectively, and obtaining test results as shown in the following table:
The multi-point data language uses the operations of calling request, result verification and the like of the main number and the rake operator digital object warehouse node broadcast data language, the execution performance of the single-point data language is better than that of the cluster data language, and the data language clusters of the 5-number Rayleigh digital object warehouse node are better than the data language clusters of the 7-number Rayleigh digital object warehouse node. The check all users and balance method takes longer because it needs to check the balance of 30000 accounts. In summary, call execution time for multi-point data languages is acceptable.
Through the above experiments and tests, the practicality of the recovery method of the present invention was confirmed to be fully feasible.
Embodiments of the present invention are described with reference to flowchart illustrations and/or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal device, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
Finally, it is further noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or terminal that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or terminal. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article or terminal device comprising the element.
The foregoing description of embodiments of the invention, wherein specific examples are employed to illustrate the principles and implementations of the invention, is provided by way of example only to assist in understanding the methods and concepts underlying the invention; meanwhile, as those skilled in the art will have variations in the specific embodiments and application scope in accordance with the ideas of the present invention, the present description should not be construed as limiting the present invention in view of the above.
Claims (10)
1. The recovery method of the memory state data for the data language is characterized in that the recovery method is applied to a plurality of digital object warehouse nodes of a digital network, each digital object warehouse node of the plurality of digital object warehouse nodes of the digital network is used for running the data language, the data language is used for a multi-point data language, the plurality of digital object warehouse nodes of the digital network form a data language cluster, and the recovery method comprises the following steps:
receiving and executing a call request by a master number and rayleigh number digital object warehouse node in the data language cluster, and broadcasting the call request to all slave number and rayleigh number digital object warehouse nodes in the data language cluster by the master number and rayleigh number digital object warehouse node;
the main digital object warehouse node receives all calling results fed back by the secondary digital object warehouse node;
under the condition that the calling result fed back by a first slave digital object warehouse node in all slave digital object warehouse nodes is wrong, the master digital object warehouse node sends full-quantity state data used by multi-point data languages running by the master digital object warehouse node to the first slave digital object warehouse node;
Under the condition that the master digital object warehouse node executes the calling request and determines that the calling result is wrong, the master digital object warehouse node sends a state data request to a second slave digital object warehouse node in all slave digital object warehouse nodes, wherein the second slave digital object warehouse node is any slave digital object warehouse node in all slave digital object warehouse nodes with correct calling results;
the master digital object warehouse node receives full-quantity state data for multi-point data languages running on the second slave digital object warehouse node, which is sent by the second slave digital object warehouse node;
in the case that the first slave digital object warehouse node is offline and online again, the first slave digital object warehouse node sends recovery information to the master digital object warehouse node, and the master digital object warehouse node sends full state data for the self-operated multi-point data language to the first slave digital object warehouse node or sends incremental state data for the self-operated multi-point data language to the first slave digital object warehouse node or does not send any data to the first slave digital object warehouse node according to the recovery information and the current operation mode of the data language cluster;
When the master number Rayleigh-end digital object warehouse node is offline and online again, and the master number Rayleigh-end digital object warehouse node is changed into a slave number Rayleigh-end digital object warehouse node, and the second slave number Rayleigh-end digital object warehouse node is changed into a new master number Rayleigh-end digital object warehouse node, the master number Rayleigh-end digital object warehouse node sends the recovery information to the second slave number Rayleigh-end digital object warehouse node, and the second slave number Rayleigh-end digital object warehouse node sends full state data for self-running multi-point data words to the master number Rayleigh-end digital object warehouse node or sends incremental state data for self-running multi-point data words to the master number Rayleigh-end digital object warehouse node or does not send any data to the master number Rayleigh-end digital object warehouse node according to the recovery information and the current running mode of the data words;
and under the condition that the main number rayleigh operator digital object warehouse node is offline and online again and the main number rayleigh operator digital object warehouse node is still the main number rayleigh operator digital object warehouse node, the main number rayleigh operator digital object warehouse node executes the following steps: and receiving and executing a call request, wherein the master digital object warehouse node broadcasts the call request to all slave digital object warehouse nodes in the data language cluster.
2. The restoration method as recited in claim 1, further comprising, before a master number rake digital object repository node in the data-language cluster receives and executes the call request:
the main number and rayleigh operator digital object warehouse node determines that the operation mode is a common operation mode under the condition that the multi-point data language is classified into stateless data language;
the main number and the plurality of sub-operator digital object warehouse nodes are classified into stateful data language uses in the multi-point data language use, and the main number and the plurality of sub-operator digital object warehouse nodes in the data language use cluster are between the total number of the sub-operator digital object warehouse nodes and the total number of the current normal operation sub-operator digital object warehouse nodes, and under the condition that a first preset condition is met, the main number and the plurality of sub-operator digital object warehouse nodes determine that the operation mode is the common operation mode;
the main number-end operator digital object warehouse node is classified into a stateful data language in the multi-point data language, and the main number-end operator digital object warehouse node determines that the operation mode is a stable operation mode under the condition that a second preset condition is met between the total number of the number-end operator digital object warehouse nodes in the data language cluster and the total number of the number-end operator digital object warehouse nodes in the current normal operation;
Under the condition of the common operation mode, each data-language cluster operates in a memory persistence mode by each data-language operator digital object warehouse node;
under the condition of the stable operation mode, each digital object warehouse node records the full-quantity state data and the increment state data for the multi-point data language into a hard disk of the node;
when the recording quantity of the incremental state data for the multi-point data language reaches a preset quantity, each digital object warehouse node of the digital object database of the data Rate respectively updates and records the full state data for the multi-point data language, and simultaneously respectively deletes the incremental state data records for the multi-point data language;
wherein, the first preset condition is: the total number of the current normal running digital object warehouse nodes of the digital speech clusters is larger than the total number of the preset operation on the total number of the digital object warehouse nodes of the digital speech clusters, and the preset operation is as follows: half the total number of the digital object warehouse nodes of the number-of-Rayleigh operator in the data language cluster to obtain a first result, and rounding the first result upwards;
the second preset condition is: the total number of the digital object warehouse nodes of the number Rayleigh operator which is currently and normally operated, and the total number is not larger than the total number after the preset operation is carried out on the total number of the digital object warehouse nodes of the number Rayleigh operator in the data language cluster.
3. The restoration method according to claim 2, wherein the restoration information includes: the operation mode of the data language cluster before the first slave digital object warehouse node is offline and the version number of the multi-point data language operated after the first slave digital object warehouse node is offline again;
in the case that the first slave digital object warehouse node is offline and online again, the first slave digital object warehouse node transmits restoration information to the master digital object warehouse node, and the master digital object warehouse node transmits full state data for a multi-point data language running by itself to the first slave digital object warehouse node or transmits incremental state data for the multi-point data language running by itself to the first slave digital object warehouse node or does not transmit any data to the first slave digital object warehouse node according to the restoration information and the current operation mode of the data language cluster, including:
when the data language cluster is in the normal operation mode before the first slave digital object warehouse node is offline, and the data language cluster is in any operation mode after the first slave digital object warehouse node is online again, the master digital object warehouse node determines whether the version number of the multi-point data language cluster operated after the first slave digital object warehouse node is online again is the same as the current state version number of the data language cluster, if the version number of the multi-point data language cluster is the same, the master digital object warehouse node does not send any data to the first slave digital object warehouse node, if the version number of the multi-point data language cluster is not the same, the master digital object warehouse node sends the full state data of the multi-point data language operated by the master digital object warehouse node to the first slave digital object warehouse node, so that the first slave digital object warehouse node operates the multi-point data language of the master digital object warehouse node based on the received full state data;
When the data language cluster is in the stable operation mode before the first slave digital object warehouse node is offline and the data language cluster is in the normal operation mode after the first slave digital object warehouse node is offline again, the master digital object warehouse node sends full-quantity state data of the self-operated multi-point data language to the first slave digital object warehouse node so that the first slave digital object warehouse node operates the self multi-point data language based on the received full-quantity state data;
under the condition that the data word cluster is in the stable operation mode before the first slave digital object warehouse node is offline, and the data word cluster is in the stable operation mode after the first slave digital object warehouse node is offline again, the first slave digital object warehouse node acquires full-quantity state data and incremental state data for the multi-point data word from a hard disk of the first slave digital object warehouse node, meanwhile, the master digital object warehouse node determines whether a version number for the multi-point data word operated after the first slave digital object warehouse node is offline again is the same as a current state version number of the data word cluster, if the version number is the same, the master digital object warehouse node does not send any data to the first slave digital object warehouse node, and if the version number is different, the master digital object warehouse node sends the new incremental state data for the multi-point data word operated by the master digital object warehouse node to the first slave digital object warehouse node, so that the first slave digital object warehouse node does not receive the full-quantity state data word incremental data for the multi-point data word based on the first slave digital object warehouse node and the full-quantity data word cluster, and the incremental state data from the first slave digital object warehouse node does not receive the full-quantity data word from the first slave digital object warehouse node.
4. The restoration method according to claim 2, wherein the restoration information further includes: the operation mode of the data language cluster before the main number and the plurality of data language version numbers are used for the operation after the main number and the plurality of data object nodes are used for the operation;
in the case where the master number rayleigh digital object warehouse node is offline and online again, and the master number rayleigh digital object warehouse node becomes a slave number rayleigh digital object warehouse node, the master number rayleigh digital object warehouse node transmits the restoration information to the second slave number rayleigh digital object warehouse node, and the second slave number rayleigh digital object warehouse node transmits full-scale state data for a self-operated multi-point data word to the master number rayleigh digital object warehouse node or transmits incremental state data for a self-operated multi-point data word to the master number rayleigh digital object warehouse node or does not transmit any data to the master number rayleigh digital object warehouse node according to the restoration information and the current operation mode of the data word cluster, including:
Under the condition that the data language cluster is in the common operation mode before the main number and the plurality of sub-operator digital object warehouse nodes are off line and the data language cluster is in any operation mode after the main number and the sub-operator digital object warehouse nodes are on line again, the second slave digital object warehouse node determines whether the version number of the multi-point data language running after the master digital object warehouse node is on line again is the same as the version number of the current state of the data language cluster, if the data are the same, the second slave digital object warehouse node does not send any data to the master digital object warehouse node, and if the data are not the same, the second slave digital object warehouse node sends the total state data for the self-operated multi-point data language to the master digital object warehouse node, so that the master digital object warehouse node operates the self-operated multi-point data language based on the received total state data;
when the data language cluster is in the stable operation mode before the master number and Rayleigh operator digital object warehouse node is offline and the data language cluster is in the normal operation mode after the master number and Rayleigh operator digital object warehouse node is offline again, the second slave number and Rayleigh operator digital object warehouse node sends full-quantity state data of the self-operated multi-point data language to the master number and Rayleigh operator digital object warehouse node so that the master number and Rayleigh operator digital object warehouse node operates the self multi-point data language based on the received full-quantity state data;
And under the condition that the data word cluster is in the stable operation mode before the master number Rayleigh operator digital object warehouse node is offline and the data word cluster is in the stable operation mode after the master number Rayleigh operator digital object warehouse node is offline again, the master number Rayleigh operator digital object warehouse node acquires full-quantity state data and incremental state data for the multi-point data word from a hard disk of the master number Rayleigh operator digital object warehouse node, and meanwhile, the second slave number Rayleigh operator digital object warehouse node determines whether the version number for the multi-point data word operated after the master number Rayleigh operator digital object warehouse node is offline again is the same as the current state version number of the data word cluster, if the version number for the multi-point data word is the same, the second slave number Rayleigh operator digital object warehouse node does not send any data to the master number Rayleigh operator digital object warehouse node, and if the version number for the multi-point data word operated by the second slave number Rayleigh operator digital object warehouse node is different, the second slave number Rayleigh operator digital object warehouse node sends the new incremental state data for the multi-point data word operated by the master number Rayleigh operator digital object warehouse node to the master number warehouse node so that the master number Rayleigh operator digital object warehouse node obtains the full-quantity data based on the received from the state data of the master number Rayleigh operator digital object warehouse node and the full-quantity data.
5. A recovery method according to claim 3, wherein the master digital object repository node determining whether the version number for the multi-point data language operated after the first slave digital object repository node is re-online is the same as the current state version number for the cluster for the data language comprises:
under the condition that a multi-point data language version number used by the first slave digital object warehouse node is not crashed, the recovery information sent to the master digital object warehouse node by the first slave digital object warehouse node carries the multi-point data language version number operated after the first slave digital object warehouse node is on line again;
the master digital object warehouse node determines whether the version number for the multi-point data language operated after the first slave digital object warehouse node is on line again is the same as the current state version number of the data language cluster or not based on the received version number for the multi-point data language operated after the first slave digital object warehouse node is on line again;
under the condition that a multi-point data language version number on the first slave digital object warehouse node breaks down, the recovery information sent to the master digital object warehouse node by the first slave digital object warehouse node does not carry the multi-point data language version number operated after the first slave digital object warehouse node is on line again;
And the master digital object warehouse node does not receive the version number for the multi-point data language operated after the first slave digital object warehouse node is on line again, and the version number for the multi-point data language operated after the first slave digital object warehouse node is on line again is determined to be different from the current state version number of the data language cluster.
6. The restoration method according to claim 2, characterized in that the restoration method further comprises:
before the first slave digital object warehouse node is offline, if the total number of the digital object warehouse nodes of the data language cluster is between the total number of the digital object warehouse nodes of the digital object and the total number of the digital object warehouse nodes of the digital object in the current normal operation, the master digital object warehouse node determines that the operation mode is the common operation mode;
after the first slave digital object warehouse node is offline, and when the first slave digital object warehouse node is not online again, if the total number of digital object warehouse nodes of the digital language cluster is between the total number of digital object warehouse nodes of the first slave digital object warehouse node and the total number of digital object warehouse nodes of the current normal operation, the master digital object warehouse node still meets the first preset condition, and the operation mode is still the normal operation mode; or alternatively
If the second preset condition is met between the total number of the digital object warehouse nodes of the number Rayleigh operator in the data language cluster and the total number of the digital object warehouse nodes of the number Rayleigh operator in the current normal operation, the main digital object warehouse node switches the operation mode into the stable operation mode;
when the first slave digital object warehouse node is offline and is not online again, and the operation mode is the normal operation mode, after the first slave digital object warehouse node is online again, if the total number of the digital object warehouse nodes of the data language cluster is between the total number of the digital object warehouse nodes of the current normal operation and the total number of the digital object warehouse nodes of the current normal operation, the master digital object warehouse node keeps the operation mode to be the normal operation mode; or alternatively
If the second preset condition is met between the total number of the digital object warehouse nodes of the number Rayleigh operator in the data language cluster and the total number of the digital object warehouse nodes of the number Rayleigh operator in the current normal operation, the main digital object warehouse node switches the operation mode into the stable operation mode;
When the first slave digital object warehouse node is offline and is not online again, and the operation mode is the stable operation mode, after the first slave digital object warehouse node is online again, if the total number of the digital object warehouse nodes of the data language cluster is between the total number of the digital object warehouse nodes of the current normal operation and the total number of the digital object warehouse nodes of the current normal operation, the master digital object warehouse node switches the operation mode into the normal operation mode; or alternatively
If the second preset condition is still met between the total number of the digital object warehouse nodes of the number of the Rayleigh operators in the data language cluster and the total number of the digital object warehouse nodes of the number Rayleigh operators in the current normal operation, the main digital object warehouse node keeps the operation mode still to be the stable operation mode; after the master number and Rayleigh operator digital object warehouse node switches the operation mode from the stable operation mode to the normal operation mode, all the number and Rayleigh operator digital object warehouse nodes in the data language cluster delete the full-quantity state data and the increment state data recorded in the hard disk during the stable operation mode.
7. The restoration method according to claim 1, wherein the master digital object repository node is offline and re-online, and the master digital object repository node becomes a slave digital object repository node, the second slave digital object repository node becomes a master digital object repository node, comprising:
under the condition that the main digital data object warehouse node is offline and the offline time length exceeds the preset time length and is not online again, other digital data object warehouse nodes in the data language cluster are disconnected with the main digital data object warehouse node, and a reelect request is sent to a digital data networking identification analysis system in the digital network;
after receiving re-election requests sent by all the digital object warehouse nodes except the main digital object warehouse node, the digital networking identification analysis system elects the second slave digital object warehouse node from all the digital object warehouse nodes except the main digital object warehouse node as the new main digital object warehouse node;
After the second slave digital object warehouse node becomes the new master digital object warehouse node, sending a connection request to all digital object warehouse nodes except the master digital object warehouse node so that all digital object warehouse nodes except the master digital object warehouse node are connected with the second slave digital object warehouse node;
and after the master digital object warehouse node is on line again, the second slave digital object warehouse node sends the connection request to the master digital object warehouse node so that the master digital object warehouse node connects the second slave digital object warehouse node with the identity of the slave digital object warehouse node.
8. The restoration method according to claim 1, wherein the primary number rayleigh digital object warehouse node is offline and online again, and the primary number rayleigh digital object warehouse node remains a primary number rayleigh digital object warehouse node, comprising:
under the condition that the main number rayleigh operator digital object warehouse node is offline and the offline time is offline again when the preset time is not exceeded, other number rayleigh operator digital object warehouse nodes in the data language cluster cannot perceive that the main number rayleigh operator digital object warehouse node is offline, and connection with the main number rayleigh operator digital object warehouse node is still maintained;
And after the primary digital object warehouse node is online again, executing the following steps: and receiving and executing a call request, wherein the master digital object warehouse node broadcasts the call request to all slave digital object warehouse nodes in the data language cluster.
9. A system for recovering memory state data for a data language, the system comprising: a plurality of digital object repository nodes of a digital network, the plurality of digital object repository nodes of the digital network comprising: a master number of rayleigh digital object warehouse nodes and a slave number of rayleigh digital object warehouse nodes; each of the plurality of digital object warehouse nodes operates a data language, the data language is a multi-point data language, the plurality of digital object warehouse nodes form a cluster for the data language, and the main digital object warehouse node comprises:
the receiving and executing broadcasting module is used for receiving and executing the calling request and broadcasting the calling request to all slave digital object warehouse nodes in the data language cluster;
The receiving result module is used for receiving all calling results fed back from the digital object warehouse node of the digital Rayleigh operator;
the first sending module is used for sending the full-quantity state data of the multi-point data language running by the first slave digital object warehouse node to the first slave digital object warehouse node under the condition that the calling result fed back by the first slave digital object warehouse node in all slave digital object warehouse nodes is wrong;
the data request module is used for sending a state data request to a second slave digital object warehouse node in all slave digital object warehouse nodes under the condition that the calling request is executed and the calling result of the data request module is determined to be wrong, wherein the second slave digital object warehouse node is any slave digital object warehouse node in all slave digital object warehouse nodes with correct calling results;
a data receiving module, configured to receive full-quantity state data for a multi-point data language running on the second slave digital object warehouse node, where the full-quantity state data is sent by the second slave digital object warehouse node;
the receiving information sending data module is used for receiving recovery information sent by the first slave digital object warehouse node when the first slave digital object warehouse node is offline and online again, sending full state data for the self-operated multi-point data language to the first slave digital object warehouse node or sending the incremental state data for the self-operated multi-point data language to the first slave digital object warehouse node or not sending any data to the first slave digital object warehouse node according to the recovery information and the current operation mode of the data language cluster;
The information sending and receiving data module is used for sending the recovery information to the second slave digital object warehouse node under the condition that the second slave digital object warehouse node is offline and online again and becomes a slave digital object warehouse node, and the second slave digital object warehouse node becomes a new master digital object warehouse node, and receiving the total state data or the incremental state data for the multi-point data words operated by the second slave digital object warehouse node according to the recovery information and the operation mode;
the execution module is used for executing the steps under the condition that the execution module is offline and online again and is still a master number and Rate operator digital object warehouse node: and receiving and executing the call request, and broadcasting the call request to all slave digital object warehouse nodes in the data language cluster.
10. A digital networking system, comprising: the system comprises a digital Rayleigh network identification analysis system, a plurality of digital object warehouse nodes stored in a digital manner and a plurality of digital object warehouse nodes stored in a digital manner;
The digital network identification analysis system is used for managing whether a plurality of digital object warehouse nodes of the digital Rayleigh operator and a plurality of digital object warehouse nodes of the digital Rayleigh operator join the digital network system;
a plurality of said digital object warehouse nodes for performing the method for recovering data-in-speech memory state data as claimed in any one of claims 1-8;
a plurality of the plurality of digital object warehouse nodes are configured to store at least one type of information including:
the operation result of the multi-point data language;
state data for the multi-point data language;
and the execution branch information for the multi-point data language.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310968475.9A CN116955013A (en) | 2023-08-03 | 2023-08-03 | Method, system and digital networking system for recovering memory state data for data language |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310968475.9A CN116955013A (en) | 2023-08-03 | 2023-08-03 | Method, system and digital networking system for recovering memory state data for data language |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116955013A true CN116955013A (en) | 2023-10-27 |
Family
ID=88447421
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310968475.9A Pending CN116955013A (en) | 2023-08-03 | 2023-08-03 | Method, system and digital networking system for recovering memory state data for data language |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116955013A (en) |
-
2023
- 2023-08-03 CN CN202310968475.9A patent/CN116955013A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106878473B (en) | Message processing method, server cluster and system | |
| CN108683516B (en) | Application instance upgrading method, device and system | |
| CN110427299B (en) | Log processing method, related device and system for micro-service system application | |
| CN110888889B (en) | Data information updating method, device and equipment | |
| CN110708196B (en) | Data processing method and device | |
| US20080288812A1 (en) | Cluster system and an error recovery method thereof | |
| CN110196761A (en) | Delay task processing method and processing device | |
| CN106452836B (en) | main node setting method and device | |
| CN105426271A (en) | Lock management method and device for distributed storage system | |
| CN104750573A (en) | Whole-consistency back-up and restoration methods for data nodes of distributed data system | |
| CN107122271B (en) | Method, device and system for recovering node event | |
| CN116701063B (en) | Persistence method, device and system for internal memory state data for data language of digital networking | |
| CN108256311B (en) | Authorization authentication method and device and electronic equipment | |
| CN114625566A (en) | Data disaster tolerance method and device, electronic equipment and storage medium | |
| CN106815318B (en) | Clustering method and system for time sequence database | |
| CN111046004B (en) | Data file storage method, device, equipment and storage medium | |
| CN107846294B (en) | Main/standby service switching method and device | |
| CN116955013A (en) | Method, system and digital networking system for recovering memory state data for data language | |
| CN116974874A (en) | Database testing method and device, electronic equipment and readable storage medium | |
| CN108829735B (en) | Synchronization method, device, server and storage medium for parallel execution plan | |
| CN102811154B (en) | Resource acquisition method and network server system | |
| CN112540875B (en) | Method for restoring check availability of mysql database based on xtrabackup | |
| CN116701410B (en) | Method and system for storing memory state data for data language of digital networking | |
| CN110086660B (en) | Data processing method and device | |
| CN108319679B (en) | Method and device for generating primary key |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |