169 results about "Coroutine" patented technology

The invention discloses a coroutine-based asynchronous system calling system, which comprises a recording module, an execution module and a return module; a calling queue is arranged between the recording module and the execution module, and a return queue is arranged between the execution module and the return module; the recording module is used for receiving a system call request initiated by acoroutine, recording system call parameters and coroutine information contained in the system call request, storing the system call parameters and the coroutine information into a call alignment, pausing the coroutine initiating system call, and then judging whether to execute system call or not; the execution module is used for executing the system call transmitted by the call queue by adoptingan auxiliary thread, and storing a system call return value and cothread information into the return queue after the system call is executed; and the return module is used for reading the system callreturn value from the return queue and adding the coroutine to the running queue according to the coroutine information so as to recover coroutine running. The asynchronous execution problem of systemcalling based on the coroutine is solved.

The invention discloses a coroutine-based data processing method and device, computer equipment and a storage medium. The method comprises the steps of receiving task set data sent by an uploading end; counting the total number of calculation tasks in the task set data to create coroutines with the same number as the total number of the tasks, and correspondingly allocating one calculation task toeach coroutine; generating a corresponding dynamic link library file according to the computing task correspondingly distributed by each coroutine; if a calling instruction of a JNI interface corresponding to the dynamic link library files is detected, starting coroutines according to the dynamic link library files respectively, operating calculation tasks in the task set data through the coroutines respectively, obtaining calculation results corresponding to the calculation tasks, and forming a calculation result set is formed; and sending the calculation result set to the uploading end. According to the method, the coroutine participates in calculation so as to reduce I / O blockage in a server, the utilization rate of a CPU in the server is also reduced, and the calculation efficiency isimproved.

The invention relates to the communication technical field, and discloses a coroutine monitoring method and apparatus for solving the technical problem of system delay caused by coroutine abnormity. The coroutine monitoring method comprises the steps of performing monitoring on at least one coroutine in a first progress through a monitoring thread in the first progress to determine whether a coroutine, the operating duration of which exceeds a preset operating duration, exists in the at least one coroutine or not, wherein the respective preset operating duration of each coroutine in the at least one coroutine is pre-recorded; and if it is determined that the operating duration of the first coroutine in the at least one coroutine exceeds a first preset operating duration set for the first coroutine, stopping the operated first coroutine through the monitoring thread.

The invention discloses a method, a device and a system for requesting data, and relates to the technical field of internets, which aims at solving the problem of paralysis caused by excessive query requests of a database. The method comprises the following steps of establishing a first coroutine corresponding to the database, detecting the load state of the database through the first coroutine, and obtaining the detection result; when an query request reported by a terminal is received, establishing a second coroutine corresponding to the query request, and obtaining the detection result from the first coroutine through the second coroutine; when the detection result represents that the database is in the overload state, stopping forwarding the query request to the database. The method is mainly applied to postgresql database systems.

The invention relates to an implementation method and system for sharing and using a Spark cluster by multiple users in a Python-Web environment. Multiple processes are introduced to support multipleSpark contexts, and the limited number of Spark contexts can be used for serving and supporting task calculation requests of multiple users. Coroutines instead of threads are used for guarding sub-processes, and the coroutines are autonomously scheduled. A task calculation result or state can be returned asynchronously so as to respond to a request of a user in real time, and the requirement of Web real-time interactive use is met. The characteristic of scheduling among Spark tasks is utilized. For some tasks with long time consumption, whether the tasks are terminated or not is given to the user, and the Web user can terminate the running Spark tasks. The problem that multiple Web users share the Spark cluster in the Python development environment can be solved.

The invention provides a microservice request response viewing method and system based on nginx. The microservice request response viewing method comprises the steps that the format of an nginx log is self-defined through log_format of the nginx, lua is embedded into the nginx through lua-nginx-module, and configuration for obtaining response data returned by a back-end server is written in a coroutine of the lua; receiving the request data and the response data by the nginx, outputting the request data and the response data by utilizing the access_log of the nginx to generate the nginx log, and outputting the nginx log to a back-end server; the configuration supporting the syslog being written in nginx.conf of the nginx, and then outputting the nginx log to a syslog server; and storing the nginx log in the syslog server in the ElasticSearch, and checking the nginx log in the ElasticSearch. The method does not affect normal operation of online micro-service applications, facilitates data query, is high in applicability, is highly decoupled from back-end applications, and is easy for later maintenance and expansion. Developers can be helped to quickly analyze and diagnose the performance bottleneck under the distributed application architecture, and the development and diagnosis efficiency under the micro-service age is improved.