CN114546644A - Cluster resource scheduling method, device, software program, electronic device and storage medium - Google Patents

Abstract

The invention provides a cluster resource scheduling method, a device, a software program, an electronic device and a storage medium, wherein the method comprises the following steps: configuring a workload in a cluster resource scheduling environment, and determining an overtime queue carrying the workload; when the work load in the overtime queue reaches an overtime state, the controller component adjusts the state of the work load into a secondary scheduling state; sending a failure copy number detection request to a corresponding estimator component based on the information of the workload; an estimator component determines a number of copies that failed scheduling in response to the failed number of copies probe request, and a scheduler component executes the cluster resource scheduler based on the number of copies that failed scheduling. Therefore, the availability of the working load can be guaranteed, the accuracy and the reliability of cluster resource scheduling are improved, the use efficiency of the cluster resources is improved, and the data processing speed of a cloud server user is ensured.

Description

Cluster resource scheduling method, device, software program, electronic device and storage medium

technical field

The present invention relates to a cluster resource scheduling processing technology of cloud networks, and in particular, to a cluster resource scheduling method, device, software program, electronic equipment and storage medium.

Background technique

With the continuous development of computer technology, cloud servers (CVM Cloud Virtual Machines) can provide safe and reliable elastic computing services, and can also provide different instance types to meet user-specific usage scenarios. These instance types consist of different combinations of CPU, memory, storage, and network. When cluster resources are scheduled during the running process of cloud servers, the speed of cluster resource scheduling directly affects the resource utilization and user experience of cloud data centers. The priority of scheduling is to ensure that users can allocate resources, and then how to allocate resources optimally, that is, to improve resource utilization. However, in the related art, during the operation of the cluster, it is still possible that the node exits abnormally, resulting in the replica having no resources for scheduling. At the same time, for the multi-cluster environment, the problem of sub-cluster resource competition may occur during task processing, which affects the It improves the accuracy and reliability of cluster resource scheduling.

SUMMARY OF THE INVENTION

In view of this, embodiments of the present invention provide a cluster resource scheduling method, device, software program, electronic device, and storage medium, which can execute the cluster resource scheduling program based on the number of replicas that fail to schedule to achieve maximum utilization in a cluster resource scheduling environment. The number of available replicas ensures the availability of workloads, improves the accuracy and reliability of cluster resource scheduling, improves cluster resource usage efficiency, ensures data processing speed for cloud server users, and improves user experience.

The technical solution of the embodiment of the present invention is realized as follows:

An embodiment of the present invention provides a cluster resource scheduling method, and the method includes:

Configuring a workload in a cluster resource scheduling environment, and determining a timeout queue to carry the workload;

When the workload in the timeout queue reaches the timeout state, the controller component adjusts the state of the workload to a secondary scheduling state;

When the scheduler component determines that the state of the workload is a secondary scheduling state, based on the information of the workload, a detection request for the number of failed replicas is sent to the corresponding estimator component;

The estimator component, in response to the failed replica number detection request, determines the number of replicas that fail to schedule, and sends the number of replicas that fail to schedule to the scheduler component;

The scheduler component executes a cluster resource scheduler based on the number of replicas that fail to schedule to achieve utilizing the cluster resource to schedule the maximum number of replicas available in the environment.

The embodiment of the present invention also provides a cluster resource scheduling device, including:

an information transmission device, configured to configure a workload in a cluster resource scheduling environment, and determine a timeout queue that carries the workload;

an information processing device, configured to adjust the state of the workload to a secondary scheduling state by the controller component when the workload in the timeout queue reaches a timeout state;

The information processing apparatus is configured to, when the scheduler component determines that the state of the workload is a secondary scheduling state, send a request for detecting the number of failed replicas to the corresponding estimator component based on the information of the workload;

the information processing apparatus, wherein the estimator component determines the number of replicas that fail to schedule in response to the failed replica number detection request, and sends the number of replicas that fail to schedule to the scheduler component;

The information processing apparatus is configured for the scheduler component to execute a cluster resource scheduler based on the number of replicas that fail to schedule, so as to realize utilizing the cluster resource to schedule the maximum number of available replicas in an environment.

In the above solution, the information processing apparatus is used for the controller component to determine the expected number of copies of the cluster resource scheduling environment;

the information processing apparatus, for the controller component to perform real-time detection on the workload based on an expected number of replicas;

The information processing apparatus is configured to adjust the workload to the timeout queue when the number of replicas in the workload is less than the expected number of replicas.

In the above solution, the information processing device is configured to, when the number of copies of the workload in the timeout queue changes, the controller component to detect the workload in the timeout queue based on the expected number of copies;

the information processing apparatus, configured to keep the workload in the timeout queue when the number of replicas in the workload is less than the expected number of replicas;

The information processing apparatus is configured to delete the workload in the timeout queue when the number of replicas in the workload is greater than or equal to the expected number of replicas.

In the above solution, the information processing apparatus is used for the estimator component to obtain node information and container group information of all nodes of the sub-cluster in the cluster resource scheduling environment;

The information processing apparatus is used for the estimator component to query the sub-cluster for a container group associated with a working copy in response to the failed copy number detection request, and to determine a container group list corresponding to the container group;

The information processing apparatus is used for the estimator component to query the container group for which scheduling fails from the container group list, and calculate the number of replicas for which scheduling fails according to the container group for which scheduling fails.

In the above solution, the information processing apparatus is configured to, when the type of the working copy is a resource type, determine a list of copy controller objects corresponding to the working copy;

the information processing apparatus, configured to search for the container group list associated with the working copy through the cache of the copy controller object list;

The information processing apparatus is configured to, when the type of the working copy is a state copy set type, search the cache of the working copy for a list of container groups associated with the working copy.

In the above solution, the information processing device is configured to, when the estimator component starts, the estimator component acquires node information and container group information of all nodes of the sub-cluster in the cluster resource scheduling environment;

the information processing apparatus, wherein the estimator component selects a node matching the workload from all the nodes in the cluster resource in response to a request for estimating the maximum number of available replicas;

The information processing device is configured to determine container group information corresponding to each node matching the workload, and based on the container group information, determine the maximum number of available copies of each node matching the workload;

The information processing apparatus is configured to determine the maximum number of available copies in the cluster resource scheduling environment based on the maximum number of available copies of each node matching the workload.

The embodiment of the present invention also provides an electronic device, the electronic device includes:

memory for storing executable instructions;

The processor is configured to implement the pre-order cluster resource scheduling method when executing the executable instructions stored in the memory.

Embodiments of the present invention further provide a computer-readable storage medium storing executable instructions, which implement a pre-order cluster resource scheduling method when the executable instructions are executed by a processor.

Embodiments of the present application provide a computer program product or computer program, where the computer program product or computer program includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. The processor of the computer device reads the computer instruction from the computer-readable storage medium, and the processor executes the computer instruction, so that the computer device executes the cluster resource scheduling method provided by the embodiment of the present application.

The embodiment of the present invention has the following beneficial effects:

In this embodiment of the present invention, a workload is configured in a cluster resource scheduling environment, and a timeout queue that carries the workload is determined; when the workload in the timeout queue reaches a timeout state, the controller component sends the workload to the workload. When the scheduler component determines that the state of the workload is a secondary scheduling state, based on the information of the workload, a detection request for the number of failed copies is sent to the corresponding estimator component; the estimation In response to the failed replica number detection request, the scheduler component determines the number of replicas that fail to schedule, and sends the number of replicas that fail to schedule to the scheduler component; the scheduler component, based on the number of replicas that fail to schedule, Execute the cluster resource scheduler to achieve the maximum number of available replicas in the scheduling environment using the cluster resources, based on the number of replicas that fail to schedule, execute the cluster resource scheduler to achieve the maximum number of available replicas in the environment using the cluster resource scheduling to ensure work It can improve the availability of load, improve the accuracy and reliability of cluster resource scheduling, improve the efficiency of cluster resource use, ensure the data processing speed of cloud server users, and improve the user experience.

Description of drawings

FIG. 1 is a schematic diagram of a usage scenario of a cluster resource scheduling method provided by an embodiment of the present invention;

2 is a schematic structural diagram of an electronic device provided by an embodiment of the present invention;

FIG. 3 is an optional schematic flowchart of a cluster resource scheduling method provided by an embodiment of the present invention;

4 is a schematic structural diagram of a cluster resource scheduling apparatus in an embodiment of the present invention;

FIG. 5 is an optional schematic flowchart of a cluster resource scheduling method provided by an embodiment of the present invention;

FIG. 6 is an optional schematic flowchart of a cluster resource scheduling method provided by an embodiment of the present invention;

FIG. 7 is an optional schematic flowchart of a cluster resource scheduling method provided by an embodiment of the present invention.

Detailed ways

In order to make the purpose, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings. All other embodiments obtained under the premise of creative work fall within the protection scope of the present invention.

In the following description, reference is made to "some embodiments" which describe a subset of all possible embodiments, but it is understood that "some embodiments" can be the same or a different subset of all possible embodiments, and Can be combined with each other without conflict.

Before further describing the embodiments of the present invention in detail, the terms and terms involved in the embodiments of the present invention are described. The terms and terms involved in the embodiments of the present invention are applicable to the following explanations.

1) In response, used to represent the condition or state on which the executed operation depends, when the dependent condition or state is satisfied, the executed one or more operations may be real-time, or may have a set delay; Unless otherwise specified, there is no restriction on the order of execution of multiple operations to be executed.

2) Terminals, including but not limited to: ordinary terminals and dedicated terminals, wherein the ordinary terminal maintains a long connection and/or a short connection with the transmission channel, and the dedicated terminal maintains a long connection with the transmission channel.

3) Client, the carrier for realizing specific functions in the terminal, for example, the mobile client (APP) is the carrier for specific functions in the mobile terminal, for example, the function of executing reports or displaying reports.

4) Component is the functional module of the view of the applet, also known as the front-end component, buttons, headers, tables, sidebars, content and footers in the page, etc. Components include modular code to facilitate Reuse in different pages of the applet.

5) Server cluster (Server cluster), refers to the collection of many servers to perform the same service together, which appears to the client as if there is only one server. A server cluster can use multiple computers for parallel computing to obtain high computing speed, and can also use multiple computers for backup, so that any one machine is broken and the entire system can still run normally. The method for handling hard disk failures in a server cluster provided in this application can be applied to cloud server usage scenarios and distributed server usage scenarios to implement status detection and fault repair of server hard disks in different usage scenarios. A server (CVM Cloud Virtual Machine) is a simple, efficient, safe and reliable computing service with scalable processing power. Its management is simpler and more efficient than the traditional single physical server. Users can quickly create or release any number of cloud servers for users' business processes without purchasing hardware in advance, and store cloud server users' data. In the distributed server use environment, the user's data and programs may not be located on one server, but distributed in multiple servers. Similarly, the use environment of the distributed server also needs to configure a large number of hard disks, and it is also necessary to pass the procedures of this application. The provided method for handling hard disk failures in a server cluster realizes state detection and fault repair of the server hard disks.

6) The container cluster management system Kubernetes, which can be called K8S, is an open source container operation platform, which can realize the functions of combining several containers into one service and dynamically assigning the host for the container to run, providing users with extremely efficient use of containers. Great convenience. Through Kubernetes, you can quickly deploy applications, quickly expand applications, seamlessly connect new application functions, and optimize the use of hardware resources.

Nodes are the basic elements of a container cluster. Nodes depend on services and can be either virtual machines or physical machines. Each node contains the basic components required to run the container group Pod, including Kubelet (container management component), Kubeproxy (network proxy component), etc.

The master node (master node) refers to the cluster control node, which manages and controls the entire cluster. All control commands of k8s are sent to it, and it is responsible for the specific execution process. The kube-apiserver (resource access component), kube-controller-mansger (running management controller component) and kube-scheduler (scheduling component) running on the Master node continuously communicate with the kubelet and kube-proxy on the worker node (Node). Communication to maintain the healthy working state of the entire cluster. If the service of the Master node cannot access a Node, the Node will be marked as unavailable, and the newly created Pod (container group) will no longer be scheduled to it. However, additional monitoring is required for the Master itself, so that the Master does not become a single point of failure of the cluster, so the Master service also needs to be deployed with high availability.

Nodes other than the Master are called Node or Worker nodes (worker nodes). You can use the node view command (kubectl get nodes) in the Master to view the Node nodes in the cluster. Each Node node will be assigned some workload (Docker container) by the Master node. When a Node goes down, the workload on the node will be automatically transferred to other nodes by the Master node.

Pod (container group): The smallest/simplest basic unit created or deployed by kubernetes---container group, a Pod represents a microservice process running on the cluster, and a microservice process encapsulates an edge that provides microservice applications Containers (there can also be multiple edge containers), storage resources, a separate network IP, and management policy options that control how the containers run.

7) Workload: A workload is a class of applications that can contain multiple replica instances.

8) Replica: The instance unit of the workload, each replica instance is an independent container.

9) Secondary scheduling: In the process of task processing, it is used to reallocate cluster resources and adapt to the scheduling of task requirements.

Before introducing the cluster resource scheduling method provided by the present application, a brief description of the defects in the related art is given first. In the related art, the following methods are usually used in the resource scheduling of the cloud network:

1) Through the overloaded node detection step, the nodes with resource utilization greater than 90% are detected as overloaded copies, and then these copies are scheduled and migrated, and finally the load balance of the entire storage cluster node can be achieved. The disadvantage of this method is that it is only applicable to a single cluster and cannot be extended to multiple clusters.

2) First, receive the application container deployment instruction sent by the user and the cluster resource usage information uploaded by the federated cluster, and then determine the number of application templates in the federated cluster through the total number of copies of the application template in the application container deployment instruction and the cluster resource usage information. The number of deployed replicas in the sub-cluster can finally realize replica scheduling that considers the operation of the sub-cluster resources. The disadvantage of this method is that the node may exit abnormally during the cluster operation, causing the replica to be scheduled without resources. There is a subcluster resource contention problem, so there is still a risk of replicas not functioning properly.

3) At regular intervals or when new nodes are added, the Pods that need to be scheduled are screened by the resource utilization of each node and the average resource utilization of all nodes in the cluster, and the Pods are migrated to the resource utilization average. The disadvantage of this method is that it cannot handle the scenario when the cluster resources are insufficient, and can only solve the cluster utilization balance when the remaining resources of a single cluster are sufficient, and the risk of system downtime will still occur.

In order to overcome the above shortcomings, the present application provides a cluster resource scheduling method, device, software program, electronic device and storage medium. FIG. 1 is a schematic diagram of a usage scenario of the cluster resource scheduling method provided by an embodiment of the present invention. With the continuous development of computer technology, cloud servers (Cloud Virtual Machine, CVM) can provide safe and reliable elastic computing services, and can also provide different instance types to meet user-specific usage scenarios. The terminals (including the terminal 10-1 and the terminal 10-2) are provided with corresponding clients capable of performing different functions, wherein the client is the terminal (including the terminal 10-1 and the terminal 10-2) from the corresponding cloud server through the network 300. 200 to obtain different information, and can deploy different services in the cloud server. The terminal is connected to the server 200 through the network 300, and the network 300 may be a wide area network or a local area network, or a combination of the two, and uses a wireless link to realize data transmission. These instance types provided by the cloud server are composed of different combinations of CPU, memory, storage and network, and store the user's business data in the hard disk of the cloud server. However, during the operation of the cloud server, a large amount of fragmentation of resources, resulting in redundancy of resources, reducing the processing speed of the cloud server network, affecting the task processing speed, and affecting the use effect of the cloud server network. In the embodiments provided by the present invention, the cloud server applications running in the cloud server 200 may be written in software code environments of different programming languages, and the code objects may be different types of code entities. For example, in C language software code, a code object can be a function. In the software code of the JAVA language, a code object can be a class, and in the OC language on the IOS side, it can be a piece of object code. In software code in C++ language, a code object can be a class or a function to execute processing instructions from different terminals. The source of the compilation environment of the cloud server is no longer distinguished in this application.

The structure of the cluster resource scheduling apparatus according to the embodiment of the present invention will be described in detail below. The cluster resource scheduling apparatus may be implemented in various forms, such as a dedicated terminal with the processing function of the cluster resource scheduling apparatus, or it may be provided with a cluster resource scheduling apparatus for processing A functional server, such as the server 200 in Figure 1 of the preamble. FIG. 2 is a schematic diagram of the composition and structure of a cluster resource scheduling apparatus provided by an embodiment of the present invention. It can be understood that FIG. 2 only shows an exemplary structure of the cluster resource scheduling apparatus, but not the entire structure. Parts shown in FIG. 2 may be implemented as needed. structure or all structures.

The cluster resource scheduling apparatus provided by the embodiment of the present invention includes: at least one processor 201, a memory 202, a user interface 203, and at least one network interface 204. Various components in the cluster resource scheduling apparatus are coupled together through the bus system 205 . It will be understood that the bus system 205 is used to implement the connection communication between these components. In addition to the data bus, the bus system 205 also includes a power bus, a control bus and a status signal bus. However, for the sake of clarity, the various buses are labeled as bus system 205 in FIG. 2 .

The user interface 203 may include a display, a keyboard, a mouse, a trackball, a click wheel, keys, buttons, a touch pad or a touch screen, and the like.

It will be appreciated that the memory 202 may be either volatile memory or non-volatile memory, and may include both volatile and non-volatile memory. The memory 202 in the embodiment of the present invention can store data to support the operation of the terminal (eg 10-1). Examples of such data include: any computer program used to operate on the terminal (eg 10-1), such as operating systems and applications. The operating system includes various system programs, such as a framework layer, a core library layer, a driver layer, etc., which are used to implement various basic services and process hardware-based tasks, and the application programs may include various application programs. The terminals involved include but are not limited to mobile phones, computers, intelligent voice interactive devices, smart home appliances, vehicle-mounted terminals, etc. The embodiments of the present invention can be applied to various scenarios, including but not limited to cloud technology, artificial intelligence, smart transportation, assisted driving, etc., to implement the cluster resource scheduling method provided by the present invention in various scenarios.

In some embodiments, the cluster resource scheduling apparatus provided by the embodiments of the present invention may be implemented by a combination of software and hardware. As an example, the cluster resource scheduling apparatus provided by the embodiments of the present invention may be processors in the form of hardware decoding processors , which is programmed to execute the cluster resource scheduling method provided by the embodiment of the present invention. For example, the processor in the form of a hardware decoding processor may use one or more Application Specific Integrated Circuits (ASIC, Application Specific Integrated Circuit), DSP, Programmable Logic Device (PLD, Programmable Logic Device), Complex Programmable Logic Device (CPLD) , Complex Programmable Logic Device), Field Programmable Gate Array (FPGA, Field-Programmable Gate Array) or other electronic components.

As an example in which the cluster resource scheduling apparatus provided by the embodiment of the present invention is implemented by combining software and hardware, the cluster resource scheduling apparatus provided by the embodiment of the present invention may be directly embodied as a combination of software modules executed by the processor 201, and the software modules may be located in a storage medium The storage medium is located in the memory 202, and the processor 201 reads the executable instructions included in the software module in the memory 202, and combines necessary hardware (for example, including the processor 201 and other components connected to the bus 205) to complete the embodiments of the present invention. The cluster resource scheduling method.

As an example, the processor 201 may be an integrated circuit chip with signal processing capabilities, such as a general-purpose processor, a digital signal processor (DSP, Digital Signal Processor), or other programmable logic devices, discrete gates or transistor logic devices , discrete hardware components, etc., where a general-purpose processor may be a microprocessor or any conventional processor, or the like.

As an example in which the cluster resource scheduling apparatus provided in the embodiment of the present invention is implemented in hardware, the apparatus provided in the embodiment of the present invention may be directly executed by a processor 201 in the form of a hardware decoding processor, for example, by one or more applications. Application Specific Integrated Circuit (ASIC, Application Specific Integrated Circuit), DSP, Programmable Logic Device (PLD, Programmable Logic Device), Complex Programmable Logic Device (CPLD, Complex Programmable Logic Device), Field Programmable Gate Array (FPGA, Field-Programmable) GateArray) or other electronic components to implement the cluster resource scheduling method provided by the embodiment of the present invention.

The memory 202 in the embodiment of the present invention is used for storing various types of data to support the operation of the cluster resource scheduling apparatus. Examples of these data include: any executable instructions for operating on the cluster resource scheduling apparatus, such as executable instructions, and the program implementing the method for scheduling slave cluster resources according to the embodiment of the present invention may be included in the executable instructions.

In other embodiments, the cluster resource scheduling apparatus provided by the embodiments of the present invention may be implemented in software. FIG. 2 shows the cluster resource scheduling apparatus stored in the memory 202, which may be software in the form of programs and plug-ins. It also includes a series of modules. As an example of the program stored in the memory 202, it may include a cluster resource scheduling apparatus. The cluster resource scheduling apparatus includes the following software modules, an information transmission module 2081 and an information processing module 2082. When the software modules in the cluster resource scheduling device are read into the RAM by the processor 201 and executed, the cluster resource scheduling method provided by the embodiment of the present invention will be implemented, wherein the functions of each software module in the cluster resource scheduling device include:

The information transmission device 2081 is configured to configure a workload in a cluster resource scheduling environment, and determine a timeout queue that carries the workload.

An information processing device 2082, configured to adjust the state of the workload to a secondary scheduling state when the workload in the timeout queue reaches a timeout state.

The information processing device 2082 is configured to, when the scheduler component determines that the state of the workload is a secondary scheduling state, send a detection request for the number of failed replicas to the corresponding estimator component based on the information of the workload.

The information processing device 2082 is configured for the estimator component to determine, in response to the failed replica number detection request, the number of replicas that fail to schedule, and send the number of replicas that fail to schedule to the scheduler component.

The information processing device 2082 is configured for the scheduler component to execute a cluster resource scheduler based on the number of replicas that fail to schedule, so as to achieve the maximum number of available replicas in the environment by utilizing the cluster resource scheduling.

The information processing device 2082 is configured to, in response to the cluster resource scheduling mode matching the cluster resource, configure the corresponding cluster resource for the to-be-processed task according to the priority of the to-be-processed task.

According to the electronic device shown in FIG. 2, in one aspect of the present application, the present application further provides a computer program product or computer program, the computer program product or computer program comprising computer instructions, the computer instructions being stored in a computer-readable in the storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the methods provided in various optional implementation manners of the foregoing cluster resource scheduling methods.

Referring to FIG. 3, FIG. 3 is a schematic flowchart of an optional cluster resource scheduling method provided by an embodiment of the present invention. It is understood that the steps shown in FIG. 3 may be performed by various electronic devices running the cluster resource scheduling apparatus. It is a control terminal such as a dedicated terminal with cluster resource scheduling function, a server or a server cluster controller, and a cloud network server. Wherein, the dedicated terminal with the cluster resource scheduling device can be packaged in the server 200 shown in FIG. 1 to execute the corresponding software modules in the cluster resource scheduling device shown in the preceding sequence of FIG. 2 . The steps shown in FIG. 3 will be described below.

Step 301: The cluster resource scheduling apparatus configures the workload in the cluster resource scheduling environment, and determines a timeout queue that carries the workload.

In some embodiments of the present invention, for a cloud server cluster environment, the cluster resource scheduling apparatus may include different types of components, for example, including: a controller component, a scheduler component, and an estimator component, specifically, referring to FIG. 4 , 4 is a schematic diagram of the architecture of a cluster resource scheduling apparatus according to an embodiment of the present invention, wherein the controller component is used to detect all workloads, and only place workloads that do not reach the expected number of replicas into a timeout queue for timing. When a timeout event occurs, the status of the workload will be updated, indicating that it needs to be rescheduled. The scheduler component is used to continuously detect all workloads. When there is a workload that needs to be rescheduled, it will send a request to the estimator to obtain the number of replicas that the workload failed to schedule in each cluster, and use the results Secondary scheduling. The estimator component is used to detect the cluster replicas and nodes of a sub-cluster to count the cluster resource usage. When receiving a failed replica statistics request from the scheduler, it will calculate and return the number of replicas that failed to schedule in the cluster in real time. The controller component, scheduler component, estimator component, and user-created workloads together form the control plane.

The embodiments of the present invention can be implemented in combination with cloud technology. Cloud technology refers to unifying a series of resources such as hardware, software, and network in a wide area network or a local area network to realize data computing, storage, processing, and sharing. Hosting technology can also be understood as a general term for network technology, information technology, integration technology, management platform technology and application technology based on cloud computing business model applications. Background services of technical network systems require a lot of computing and storage resources, such as video websites, picture websites and more portal websites, so cloud technology needs to be supported by cloud computing.

It should be noted that cloud computing is a computing model that distributes computing tasks on a resource pool composed of a large number of computers, enabling various application systems to obtain computing power, storage space and information services as needed. The network that provides the resources is called the "cloud". The resources in the "cloud" are infinitely expandable in the eyes of users, and can be obtained at any time, used on demand, expanded at any time, and paid according to usage. As a basic capability provider of cloud computing, it will establish a cloud computing resource pool platform, referred to as a cloud platform, generally referred to as Infrastructure as a Service (IaaS, Infrastructure as a Service), deploying various types of virtual resources in the resource pool for External customers choose to use. The cloud computing resource pool mainly includes: computing devices (which can be virtualized machines, including operating systems), storage devices, and network devices. When a user uses a cloud server to store data or deploy different application processes, the operating parameters of the server cluster hard disk can be detected, and possible server cluster hard disk failures can be detected in time to avoid user data loss due to server cluster hard disk failures with failure warnings.

Cloud storage is a new concept extended and developed from the concept of cloud computing. Distributed cloud storage system (hereinafter referred to as storage system) refers to functions such as cluster application, grid technology and distributed storage file system. A storage system that integrates a large number of different types of storage devices (also called storage nodes) in the network through application software or application interfaces to work together to provide external data storage and service access functions. At present, the storage method of the storage system is as follows: creating a logical volume, and when creating a logical volume, a physical storage space is allocated to each logical volume, and the physical storage space may be composed of a storage device or disks of several storage devices. The client stores data on a logical volume, that is, stores the data on the file system. The file system divides the data into many parts, each part is an object, and the object contains not only data but also data identification (ID, ID entity) and other additional information, the file system writes each object into the physical storage space of the logical volume, and the file system records the storage location information of each object, so that when the client requests to access data, the file system can The storage location information of the object allows the client to access the data. The process of allocating physical storage space by the storage system to the logical volume, specifically: according to the capacity estimation of the objects stored in the logical volume (this estimation often has a large margin relative to the actual capacity of the objects to be stored) and independent redundant disks Array (RAID, Redundant Array of Independent Disk) group, which divides the physical storage space into stripes in advance, and a logical volume can be understood as a stripe, thereby allocating physical storage space for the logical volume.

Taking the implementation of the cluster resource scheduling method provided by this application through a cloud server network as an example, by configuring a workload in a cluster resource scheduling environment, and determining a timeout queue that carries the workload; When the load reaches the timeout state, the controller component adjusts the state of the workload to the secondary scheduling state; when the scheduler component determines that the state of the workload is the secondary scheduling state, based on the information of the workload, the corresponding The estimator component sends a detection request for the number of failed replicas; the estimator component determines the number of replicas that fail to schedule in response to the detection request for the number of failed replicas, and sends the number of replicas that fail to schedule to the scheduler component; The scheduler component executes a cluster resource scheduler based on the number of replicas that fail to schedule, so as to utilize the maximum number of available replicas in the cluster resource scheduling environment, thereby ensuring the availability of workloads and improving the efficiency of cluster resource scheduling. Accuracy and reliability, improve the efficiency of cluster resource usage.

When applied to a cloud product, the front end of the cloud product may be a Web UI component, which is used to receive Spark-related parameters filled in by the user, and generate job data according to the Spark-related parameters. The cluster manager (ClusterManager) may be an open source cluster resource scheduling platform such as YARN, Mesos, or Kubernetes. Spark itself already supports these open source platforms, that is, the protocols between Spark components and ClusterManager components are compatible. Driver is a job driver, Work Node is a worker node, Executor is a task execution component, and task is the smallest execution unit. Further, the structured data package (spark SQL) is a package used by Spark to operate structured data. Through the Spark SQL, the data can be queried using the SQL language. The Spark SQL supports multiple data sources, such as data warehouses. Tools (Hive) table, etc. The stream computing component is a component provided by Spark to perform stream computing on real-time data, and provides an API Application Programming Interface for operating the data stream.

Step 302: When the workload in the timeout queue reaches the timeout state, the controller component adjusts the state of the workload to a secondary scheduling state.

In some embodiments of the present invention, configuring a workload in a cluster resource scheduling environment, and determining a timeout queue to carry the workload, includes:

The controller component determines the expected number of copies of the cluster resource scheduling environment; the controller component performs real-time detection on the workload based on the expected number of copies; when the number of copies in the workload is less than the expected number of copies When the number is reached, adjust the workload to the timeout queue. Because the cloud server cluster is used in a variety of environments, the value of the expected number of copies can be flexibly set according to the use environment of the cloud server cluster, for example, the instant messaging client Tenpay payment or the instant messaging client When processing the information of borrowing funds to purchase items in , due to the large number of tasks, the expected number of copies can be set to 10,000; for video processing tasks that can be completed only through a single server cluster, the expected number of copies can be set to 100, so as to make full use of the resources of the server cluster and reduce the resource waste of the server cluster.

In some embodiments of the present invention, when the number of replicas of the workload in the timeout queue changes, the controller component detects the workload in the timeout queue based on the expected number of replicas; when the When the number of replicas in the workload is less than the expected number of replicas, the workload is kept in the timeout queue; when the number of replicas in the workload is greater than or equal to the expected number of replicas, the timeout queue is placed Workload deletion in . Since the number of copies of workloads in the timeout queue is dynamically changing data, by real-time detection of the workloads in the timeout queue, the workloads in the timeout queue can be adjusted in time and the workload for secondary resource adjustment can be reduced. quantity.

Step 303 : when the scheduler component of the cluster resource scheduling apparatus determines that the state of the workload is a secondary scheduling state, based on the information of the workload, a request for detecting the number of failed replicas is sent to the corresponding estimator component.

The workload information may include resources such as StatefulSet, Deployment, ReplicaSet, and Daemonset. The resource information includes the number of application instances and the affinity rules of the application instances. Only application instances that are compatible with Workload's affinity rules can be deployed on this computing node. A resource object in a Kubernetes cluster can be an application (APP) in a Kubernetes cluster, for example, deployment (Deployment), state replica set (StatefulSet), and routing (Ingress), container group (pod), container (container), service ( One or more of resources such as Service), Replication Controller (RC, ReplicationController).

Step 304: The estimator component of the cluster resource scheduling apparatus, in response to the failed replica number detection request, determines the number of replicas that fail to schedule, and sends the number of replicas that fail to schedule to the scheduler component.

The working process of determining the number of replicas that fail to be scheduled is further described below with reference to FIG. 5 .

Referring to FIG. 5, FIG. 5 is a schematic flowchart of an optional cluster resource scheduling method provided by an embodiment of the present invention. It is understood that the steps shown in FIG. 5 may be performed by various electronic devices running the cluster resource scheduling apparatus, for example, It is a control terminal such as a dedicated terminal with cluster resource scheduling function, a server or a server cluster controller, and a cloud network server. Wherein, the dedicated terminal with the cluster resource scheduling device can be packaged in the server 200 shown in FIG. 1 to execute the corresponding software modules in the cluster resource scheduling device shown in the preceding sequence of FIG. 2 . The steps shown in FIG. 5 will be described below.

Step 501: The estimator component obtains node information and container group information of all nodes of the sub-cluster in the cluster resource scheduling environment.

Step 502: The estimator component, in response to the failed copy number detection request, queries the sub-cluster for a container group associated with a working copy, and determines a container group list corresponding to the container group.

In some embodiments of the present invention, when the type of the working copy is a resource type, a list of copy controller objects corresponding to the working copy is determined; and the cache of the copy controller object list is used to search for the location of the working copy. A list of associated container groups. Taking K8S as an example, a Kubernetes cluster generally includes a master node (Master) and multiple computing nodes (Nodes) that communicate with the master node respectively. The master node is used to manage and control multiple A computing node, as a workload node, contains the original application directly deployed in the node and multiple container groups (Pods), each of which is encapsulated with one or more containers (Containers) for carrying applications. ), Pod is the basic operating unit of Kubernetes, the smallest deployment unit that can be created, debugged, and managed. The type of working copy is resource type (Deployment type), which can deploy type tasks. Deployment integrates online deployment, rolling upgrade, creating copies, suspending online tasks, resuming online tasks, and rolling back to a previous version (successful/stable). Deployment and other functions, to a certain extent, Deployment can achieve unattended online, which greatly reduces the complex communication and operational risks of the online process. For the working copy of the Deployment type, you can first determine the list of ReplicaSet objects associated with the Deployment type, and then pass the The replica controller ReplicaSet finds the associated Pod list from the cache. Among them, ReplicaSet is a replica controller in kubernetes. Its main function is to control the pods managed by ReplicaSet, so that the number of pod replicas is always maintained at the preset number. .

In some embodiments of the present invention, when the type of the working copy is a stateful copy set type, a list of container groups associated with the working copy is searched in the cache of the working copy, wherein for the work of the StatefulSet type For replicas, you can directly find the list of Pod objects associated with the StatefulSet type of working replicas from the cache to save search time and prompt resource scheduling speed.

Step 503: The estimator component searches the container group list for the container groups that fail to schedule, and calculates the number of replicas that fail to schedule according to the container groups that fail to schedule.

Step 305 : The scheduler component of the cluster resource scheduling apparatus executes a cluster resource scheduler based on the number of replicas that fail to schedule, so as to achieve the maximum available number of replicas in the environment using the cluster resource scheduling.

In some embodiments of the present invention, before executing the cluster resource scheduler, the maximum number of available replicas needs to be determined. Specifically, referring to FIG. 6 , FIG. 6 is an optional cluster resource scheduling method provided by the embodiments of the present invention. A schematic flowchart, it can be understood that the steps shown in FIG. 6 can be executed by various electronic devices running the cluster resource scheduling apparatus, such as a dedicated terminal with a cluster resource scheduling function, a server or a server cluster controller, a cloud network The server's controlling terminal. Wherein, the dedicated terminal with the cluster resource scheduling device can be packaged in the server 200 shown in FIG. 1 to execute the corresponding software modules in the cluster resource scheduling device shown in the preceding sequence of FIG. 2 . The steps shown in FIG. 6 will be described below.

Step 601: When the estimator component is started, the estimator component acquires node information and container group information of all nodes of the sub-cluster in the cluster resource scheduling environment.

Step 602: The estimator component selects a node matching the workload from all the nodes in the cluster resource in response to the maximum available replica quantity estimation request.

Step 603: Determine container group information corresponding to each node matching the workload, and determine the maximum number of available replicas of each node matching the workload based on the container group information.

Step 604: Determine the maximum number of available copies in the cluster resource scheduling environment based on the maximum number of copies available for each node that matches the workload.

The following describes the cluster resource scheduling method involved in the present invention by taking the cluster resource manager as the resource manager of the WeChat server as an example, wherein, in conjunction with the schematic diagram of the usage environment of the cluster resource scheduling method according to the embodiment of the present invention shown in FIG. 1 ; The terminals (including the terminal 11-1 and the terminal 11-2) are provided with corresponding clients capable of performing different functions, wherein the client is the terminal (including the terminal 11-1 and the terminal 11-2) from the corresponding server through the network 300 In 200, different information is obtained through the WeChat application for browsing, and the terminal is connected to the server 200 through the network 300. The network 300 can be a wide area network or a local area network, or a combination of the two, using a wireless link to realize data transmission, wherein, the server 200 runs The cluster resource manager matched with the WeChat application realizes resource scheduling, and the terminals (such as the terminal 10-1 and the terminal 10-2 in FIG. 1) can also be provided with a client terminal capable of displaying the corresponding software for financial lending , such as virtual resources or physical resources for financial activities or through virtual resource lending clients or plug-ins, users can obtain loans from financial institutions or platforms through the corresponding client (such as instant messaging client Tenpay payment or instant messaging client The terminal is connected to the server 200 through the network 300, and the network 300 can be a wide area network or a local area network, or a combination of the two, using a wireless link to achieve data transmission. A server (for example, the server 300 in FIG. 1 ) is a server of an enterprise that provides financial services such as payment, lending, and wealth management, such as banks, securities, and mutual funds. When the user who needs to handle the relevant financial business uses the client device to access the service provided by the client server of the enterprise, the client server can issue a payment task by triggering the applet in the instant messaging client of the user terminal. Set the expected number of replicas to 10,000. When the server cluster processes these payment tasks, in order to avoid the problem of sub-cluster resource competition, the accuracy and reliability of cluster resource scheduling are affected. Refer to Figure 7. Figure 7 is An optional schematic flowchart of the cluster resource scheduling method provided by the embodiment of the present invention, the architecture of the cluster resource scheduling is shown in FIG. 4 , and the steps shown in FIG. 7 are described below.

Step 701: Create a workload creation, and the controller component continuously judges whether the workload reaches the expected number of replicas.

Step 702: When the workload of the expected number of replicas is not reached, the controller component stores the workload in the timeout queue.

Step 703: When a replica number update event occurs for the workload in the timeout queue, the controller component determines whether the expected replica number is reached, and if it has reached, it is deleted from the timeout queue, otherwise, it is re-added to the timeout queue.

Step 704: When a workload in the timeout queue triggers a timeout, the controller component adjusts the state of the workload to a secondary scheduling state, and writes it into the workload.

Step 705: The scheduler component detects that the state of the workload is a secondary scheduling state.

Step 706: The scheduler sends a detection request for the number of failed replicas to the sub-cluster estimator according to the workload information.

Step 707: The estimator component continuously detects cluster replicas and nodes to count cluster resource usage.

Step 708: The estimator component calculates the total number of replicas that fail to schedule according to the request, and returns it to the scheduler component.

Step 709: The scheduler re-executes the scheduler according to the total number of replicas that fail to schedule, and generates a secondary scheduling result.

In some embodiments of the present invention, as shown in FIG. 4 , when the current workload A triggers secondary scheduling and the replica scheduling fails in sub-cluster 1, workload A is in sub-cluster 1, sub-cluster 2, sub-cluster 2, and sub-cluster 1. The expected number of replicas in cluster 3 is r1, r2, and r3, respectively, the number of replicas that fail to schedule in sub-cluster 1 is f1, and the maximum available replicas in sub-cluster 2 and sub-cluster 3 are m2 and m3, respectively.

In the process of generating the secondary scheduling result, when m2+m3>=f1, the scheduler component fixes the expected number of copies of workload A in subcluster 1 to r1-f1, and then uses subcluster 2 and subcluster 3 as candidate clusters, The f1 failed replicas are allocated according to the maximum available replica number ratio m2:m3, as the secondary scheduling result of the failed replicas.

In the process of generating the secondary scheduling result, when m2+m3<f1, it is determined that the secondary scheduling fails this time, and the next scheduling is waited until the secondary scheduling succeeds.

The present invention has the following beneficial technical effects:

The present invention configures the workload in the cluster resource scheduling environment, and determines the timeout queue that carries the workload; when the workload in the timeout queue reaches the timeout state, the controller component changes the status of the workload Adjust to the secondary scheduling state; when the scheduler component determines that the state of the workload is the secondary scheduling state, based on the information of the workload, it sends a detection request for the number of failed copies to the corresponding estimator component; the estimator component In response to the failed replica number detection request, determine the number of replicas that fail to schedule, and send the number of replicas that fail to schedule to the scheduler component; the scheduler component executes clustering based on the number of replicas that fail to schedule The resource scheduler can implement the maximum number of available copies in the scheduling environment using the cluster resources, and can execute the cluster resource scheduler based on the number of copies that fail to schedule to achieve the maximum number of copies available in the environment using the cluster resources. At the same time, it improves the accuracy and reliability of cluster resource scheduling, improves the utilization efficiency of cluster resources, ensures the data processing speed of cloud server users, and improves the user experience.

The above descriptions are merely examples of the present invention, and are not intended to limit the protection scope of the present invention. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present invention shall be included in the present invention. within the scope of protection.

Claims (10)

1. A method for scheduling cluster resources, the method comprising:

configuring a workload in a cluster resource scheduling environment, and determining a timeout queue carrying the workload;

when the workload in the overtime queue reaches an overtime state, a controller component adjusts the state of the workload to a secondary scheduling state;

when the scheduler component determines that the state of the workload is a secondary scheduling state, sending a detection request of the number of failed copies to a corresponding estimator component based on the information of the workload;

the estimator component responds to the detection request of the number of the failed copies, determines the number of copies failed in scheduling, and sends the number of copies failed in scheduling to the scheduler component;

the scheduler component executes a cluster resource scheduler to achieve a maximum number of available copies in the scheduling environment utilizing the cluster resources based on the number of copies that failed in the scheduling.

2. The method of claim 1, wherein configuring a workload in a cluster resource scheduling environment and determining a timeout queue carrying the workload comprises:

the controller component determining an expected number of copies of the cluster resource scheduling environment;

the controller component detects the workload in real time based on the expected copy number;

when the number of copies in the workload is less than the expected number of copies, adjusting the workload to the timeout queue.

3. The method of claim 2, further comprising:

when the number of copies of the workload in the timeout queue changes, the controller component detects the workload in the timeout queue based on the expected number of copies;

maintaining the workload in the timeout queue when the number of replicas in the workload is less than the desired number of replicas;

and deleting the workload in the timeout queue when the number of copies in the workload is greater than or equal to the expected number of copies.

4. The method of claim 1, wherein the estimator component determines a number of copies that failed scheduling in response to the probe request of the number of failed copies comprises:

the estimator component acquires node information and container group information of all nodes of a sub-cluster in a cluster resource scheduling environment;

the estimator component responds to the detection request of the number of the failed copies, inquires a container group associated with the working copy in the sub-cluster, and determines a container group list corresponding to the container group;

and the estimator component inquires the container group with failed scheduling from the container group list and calculates the number of copies with failed scheduling according to the container group with failed scheduling.

5. The method of claim 4, wherein the estimator component, in response to the failed copy number detection request, queries a container group associated with a working copy in the sub-cluster and determines a container group list corresponding to the container group, comprising:

when the type of the working copy is a resource type, determining a copy controller object list corresponding to the working copy;

searching a container group list associated with the working copy through the cache of the copy controller object list;

and when the type of the working copy is a state copy set type, searching a container group list associated with the working copy in a cache of the working copy.

6. The method of claim 1, further comprising:

when the estimator component is started, the estimator component acquires node information and container group information of all nodes of a sub-cluster in a cluster resource scheduling environment;

the estimator component filters nodes matching the workload from all nodes of the cluster resource in response to a maximum available copy number pre-estimation request;

determining container group information corresponding to each node matched with the workload, and determining the maximum available copy number of each node matched with the workload based on the container group information;

and determining the maximum available copy number in the cluster resource scheduling environment based on the maximum available copy number of each node matched with the workload.

7. An apparatus for cluster resource scheduling, the apparatus comprising:

the information transmission device is used for configuring the workload in a cluster resource scheduling environment and determining an overtime queue for bearing the workload;

the information processing device is used for adjusting the state of the workload to be in a secondary scheduling state by the controller component when the workload in the overtime queue reaches the overtime state;

the information processing device is used for sending a detection request of the number of failed copies to the corresponding estimator component based on the information of the workload when the scheduler component determines that the state of the workload is a secondary scheduling state;

the information processing device is used for the estimator component to respond to the detection request of the number of the failed copies, determine the number of copies which fail in scheduling and send the number of copies which fail in scheduling to the scheduler component;

the information processing apparatus is configured to execute, by the scheduler component, a cluster resource scheduler to achieve utilization of a maximum available number of copies in the cluster resource scheduling environment based on the number of copies that failed in the scheduling.

8. A software program, characterized in that the software program comprises:

a memory for storing executable instructions;

a processor, configured to execute the executable instructions stored in the memory, and implement the cluster resource scheduling method according to any one of claims 1 to 6.

9. An electronic device, characterized in that the electronic device comprises:

a memory for storing executable instructions;

a processor, configured to execute the executable instructions stored in the memory, and implement the cluster resource scheduling method according to any one of claims 1 to 6.

10. A computer readable storage medium storing executable instructions, wherein the executable instructions when executed by a processor implement the cluster resource scheduling method of any one of claims 1 to 6.

Images