30results about How to "Improve high availability" patented technology

The invention discloses a method for achieving a client of an access object storage cluster based on fuse. The method specifically comprises the following steps: hanging the client through a fuse kernel provided by linux, accessing a remote distributed type object storage system under the catalog where the client is hung, and processing system call formed by operation under the catalog hanging, wherein the client is divided into two parts, namely, a kernel module and a user state daemon progress module, the kernel module is based on fuse, the user state daemon progress is responsible for receiving system scheduling requests submitted by the kernel, at the same time processing the requests and feeding back the requests to the system call. Compared with the prior art, the method for achieving the client of the access object storage cluster based on fuse can greatly improve the production efficiency, improve the high availability of the cluster, is relatively high in practicability, can economize manpower, and is wide in application range and easy to popularize.

The invention discloses a gray release method capable of realizing dynamic configuration. The method specifically comprises the following steps: (1) generating a timemask timing task when an nginx server is started, and acquiring a latest gray scale strategy from a gray scale management system by the task; (2) caching the latest gray scale strategy into a Lua gray scale engine; (3) enabling the user to initiate a request through the client; (4) matching the request data of the user with the latest gray scale strategy in the lua gray scale engine; (5) if the request data is not matched with thelatest gray scale strategy, forwarding the latest gray scale strategy to an original stable version server; and (6) if the request data is not matched with the latest gray scale strategy, forwardingthe the latest gray scale strategy to a gray version server. The method has the beneficial effects that the release range is controllable, and the online risk can be reduced; flow control and cut-in are easy to realize; when a problem occurs, rolling-back operation can be carried out immediately; a target user is designated, targeted product function verification can be achieved, and effective feedback is collected; inconvenience brought to a user by service stopping can be avoided, and high availability is improved.

The invention discloses an OMCI function virtualization system of OLT equipment. A VOMCI access instance, a VOMCI function instance and an ONU state and command database are deployed in the cloud platform. The OLT sends the ONU management request to the VOMCI access instance; the VOMCI access instance is processed according to the type of the ONU management request; if the request is an ONU objectupdating request, the ONU state and command database are directly updated, and if the request is an ONU service management request, a proper VOMCI function instance is selected and the ONU service management request is forwarded to the VOMCI function instance for processing; when the VOMCI function instance processes an ONU service management request, whether the latest state data of the ONU is in a local memory or not is checked; and if the latest state data of the ONU is not in the local memory, the latest state data of the ONU is searched and downloaded from the ONU state and command database, and the latest state data of the ONU is synchronized to the ONU state and command database by the VOMCI function instance in the process of processing the ONU service management request. According to the invention, the high availability, load balance and continuous processing capability of an OLT virtualization system can be significantly improved.

The invention discloses a data processing method and device applied to databases as well as a data storage system. The method comprises steps as follows: partition indexes in a distributed cache are inquired according to inquiry parameters in an inquiry request, and partitioning keys corresponding to the inquiry parameters are acquired; if no partitioning keys are acquired from the distributed cache, partition indexes in an index database are inquired according to the inquiry parameters, and the partitioning keys are acquired; databases are accessed according to the partitioning keys. With the adoption of the data processing method and device, performance bottlenecks and single-point risks caused by a single index database can be effectively avoided, and the high usability of the system is improved.

The invention provides a method for constructing a PostgreSQL database cluster, which comprises the following steps of initializing an uninitialized node, and adding the node into a candidate node list of a main node; when only one node exists in the candidate node list of the main node, setting the node as the main node and setting the Leader Key parameter of the cluster as the node ID of the node; enabling the main node to start a PostgreSQL instance and initialize a database; after detecting the Leader Key of the cluster, enabling the slave node to start the PostgreSQL instance, and clone the database from the master node into the slave node; enabling other nodes to start a PostgreSQL instance after waiting for a Leader Key in a set cluster, and cloning a database from a master node tobecome slave nodes.

The invention relates to a high-reliability data backup and recovery method based on a cluster architecture, and its technical characteristics are: it includes an online data backup and recovery method and a historical data backup and recovery method, and the online data backup and recovery method includes: creating a table Space and divide the range partition table by time; realize a backup function during the process of importing data into the database; write the backup information into the system table; when the data of the day becomes historical data, it will be backed up according to the historical data backup method; In case of failure, reload data to realize online data recovery. The present invention adopts different backup and recovery methods for the two types of data, so that when a medium failure occurs, the high availability and high reliability of the single database node are improved, and the node cannot be effectively restored in the general cluster dual-machine hot backup scheme. Problems such as data loss when media failures occur on data and mutual backup nodes at the same time.

The present application proposes a fault handling method, device, server, and storage medium. The method is applied to the server, including: detecting whether a master database fails; if so, determining a new master database from at least one slave database, and performing a failover of the master database; obtain the routing configuration file; modify the routing information in the routing configuration file by calling the routing configuration tool through the first middleware, and obtain the updated routing configuration file; if the second middleware detects the routing configuration The routing information in the file is changed, the updated routing configuration file is reloaded, and the data connection between the second middleware and the new primary database is re-established. This application uses a self-developed routing configuration tool to automatically modify the routing information of the read-write separation middleware, which not only improves the high availability of the database, but also reduces manual intervention, saves labor costs, and realizes rapid self-healing of business traffic.

Embodiments of the present invention provide a communication method, device, electronic device, and computer storage medium. The communication method includes: determining attributes of a first network port of a virtual network component, where the virtual network component is a component for adopting NFV; When the attribute of the first network port satisfies the judgment condition, the data packet received by the first network port is forwarded through the second network port of the virtual network component; wherein, the first network port and the second network port are the virtual A pair of virtual network ports used to represent a communication link in a network component. In this way, when the attribute of the first network port satisfies the judgment condition, the data packet can be forwarded through the corresponding second network port, and the communication connection can be quickly restored when the virtual network component fails. The high availability of components can reduce the waste of resources. In addition, the virtual network components that can be applied to various deployment modes have wide applicability.

The invention discloses a method for implementing dynamic index analysis and evaluation based on micro-services, comprising: step 1), a server side builds a high-availability and high-performance micro-service cluster based on Spring Cloud; step 2), the server side receives concurrent data from users Request, the request is directly distributed to different server nodes in the microservice cluster to perform basic data verification; step 3), the valid user data received by a single node, cache the middleware redis counter + 1, and judge real-time calculation or asynchronous execution according to the machine load threshold; call The unified evaluation calculation engine immediately and asynchronously notifies the client of the result obtained from the calculation; step 4), the server sends a notification to the user to view the evaluation report through a short message and / or a system message. The method of the invention can well solve the configurability of different calculation methods for hundreds of diseases, realize online configuration, update models in time, high calculation efficiency and stable system operation.

The invention discloses an OMCI function virtualization system of an OLT device. A VOMCI access instance, a VOMCI function instance, and an ONU status and command database are deployed on a cloud platform. The OLT sends an ONU management request to the VOMCI access instance, and the VOMCI access instance Process according to the type of ONU management request, if it is an ONU object update request, directly update the ONU status and command database, if it is an ONU service management request, select a suitable VOMCI function instance and forward the ONU service management request to the VOMCI function Instance processing; when a VOMCI function instance processes an ONU service management request, check whether the latest status data of the ONU is in the local memory, if not, search and download the latest status data of the ONU from the ONU status and command database, and process the ONU business During the management request process, the VOMCI function instance synchronizes the latest status data of the ONU to the ONU status and command database. The invention can significantly improve the high availability, load balancing and connection processing capabilities of the OLT virtualized system.