CN106412016A - Connection control method and device for server - Google Patents

Abstract

The invention discloses a connection control method of a server, comprising the steps of: after stopping forwarding of service requests to service instances, forwarding a preset number of access requests to the service instances; after the service instances correctly respond to the forwarded requests , resume forwarding the service request to the service instance; when the service instance does not respond to the forwarded request, keep stopping forwarding the service request to the service instance. The invention also discloses a server connection control device. The present invention makes full use of resources and improves service experience.

Description

Server connection control method and device

technical field

The invention relates to the technical field of servers, in particular to a server connection control method and device.

Background technique

With the development and diversification of services, the number of service requests initiated by clients also increases. When there is a problem with the resources that the business service depends on, some service instances respond slowly, freeze or even fail to respond. Currently, for the above problems, business requests will continue to be forwarded to the service instance until the service instance goes down. Therefore, the current connection establishment between the client and the server cannot solve the slow response of the service instance in time, resulting in the untimely response to the business request, resulting in poor business experience.

The above content is only used to assist in understanding the technical solution of the present invention, and does not mean that the above content is admitted as prior art.

Contents of the invention

The main purpose of the present invention is to provide a server connection control method and device, aiming at solving the problem that the connection establishment between the client and the server cannot solve the slow response of the service instance in time, resulting in untimely response to business requests and poor business experience .

In order to achieve the above object, the present invention provides a server connection control method, comprising steps:

After stopping forwarding the service request to the service instance, forwarding a preset number of request accesses to the service instance;

When the service instance correctly responds to the forwarded request, resume forwarding the service request to the service instance;

When the service instance does not respond to the forwarded request, keep stopping forwarding the service request to the service instance.

Preferably, after forwarding the preset number of access requests to the service instance, it further includes:

Determine the number of requests to respond to;

Calculate the ratio of the determined quantity to the preset quantity;

When the calculated ratio is greater than the preset ratio threshold, it is determined that the service instance can correctly respond to the forwarded request.

Preferably, the method also includes:

Obtain link information of a service instance within a detection cycle;

judging whether the link information satisfies a preset disconnection condition;

Stop forwarding the service request to the service instance after the link information satisfies the preset disconnection condition.

Preferably, the judging whether the link information satisfies a preset disconnection condition includes:

Extracting the number of connection error requests from the connection information;

When the number of link errors is greater than a preset first number threshold, it is determined whether the link information satisfies a preset disconnection condition.

Preferably, the judging whether the link information satisfies a preset disconnection condition includes:

Extracting the number of connection timeout requests from the connection information;

When the number of link timeouts is greater than a preset second number threshold, it is determined whether the link information satisfies a preset disconnection condition.

In addition, in order to achieve the above object, the present invention also provides a server connection control device, including:

A control module, configured to forward a preset number of access requests to the service instance after stopping forwarding the service request to the service instance;

A recovery module, configured to recover and forward the service request to the service instance when the service instance correctly responds to the forwarded request;

The control module is further configured to stop forwarding the service request to the service instance when the service instance does not respond to the forwarded request.

Preferably, the device also includes:

A determining module, configured to determine the number of requests to respond to;

A calculation module, used to calculate the proportion of the determined quantity in the preset quantity;

A judging module, configured to judge that the service instance can correctly respond to the forwarded request when the calculated ratio is greater than a preset ratio threshold.

Preferably, the device also includes:

An acquisition module, configured to acquire link information of a service instance within a detection cycle;

A judging module, configured to judge whether the link information satisfies a preset disconnection condition;

The control module is further configured to stop forwarding the service request to the service instance after the link information satisfies a preset disconnection condition.

Preferably, the judging module includes:

A judging unit, configured to extract the number of connection error requests from the connection information;

A judging unit, configured to judge whether the link information satisfies a preset disconnection condition when the number of link errors is greater than a preset first number threshold.

Preferably, the judging unit is further configured to extract the number of link timeout requests from the link information;

The determining unit is further configured to determine whether the link information satisfies a preset disconnection condition when the number of link timeouts is greater than a preset second number threshold.

The present invention tests the disconnected service instance, and quickly recovers and forwards the service request to the service instance after the service instance correctly responds to the forwarded request, fully utilizes resources, and improves service experience.

Description of drawings

FIG. 1 is a schematic flowchart of a first embodiment of a server connection control method according to the present invention;

Fig. 2 is a schematic flow diagram of judging whether a service instance correctly responds to a request in an embodiment of the present invention;

FIG. 3 is a schematic flowchart of a second embodiment of a server connection control method according to the present invention;

FIG. 4 is a schematic flowchart of judging whether the link information meets a preset disconnection condition in an embodiment of the present invention;

FIG. 5 is a schematic flow diagram of judging whether the link information meets a preset disconnection condition in another embodiment of the present invention;

Fig. 6 is a schematic diagram of disconnection calculation in an embodiment of the present invention;

7 is a schematic flowchart of a third embodiment of a server connection control method according to the present invention;

FIG. 8 is a schematic diagram of the architecture of the client and server service requests in an embodiment of the present invention;

FIG. 9 is a schematic diagram of service instance disconnection in an embodiment of the present invention;

FIG. 10 is a schematic diagram of functional modules of the first embodiment of the connection control device of the server of the present invention;

FIG. 11 is a schematic diagram of functional modules of the second embodiment of the connection control device of the server of the present invention;

FIG. 12 is a schematic diagram of a detailed functional module of an embodiment of the judging module in FIG. 10 .

The realization of the purpose of the present invention, functional characteristics and advantages will be further described in conjunction with the embodiments and with reference to the accompanying drawings.

detailed description

The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

Referring to FIG. 1 , FIG. 1 is a schematic flowchart of a first embodiment of a server connection control method according to the present invention.

In an embodiment, the connection control method of the server includes:

Step S10, after stopping forwarding the service request to the service instance, forward a preset number of requested accesses to the service instance; after the service instance meets the disconnection condition and disconnects, the disconnection status of the service instance needs to be monitored in real time , to use the service instance for business services as soon as possible. The set time can be 2 minutes or 3 minutes, etc. After the interval is set, forward a preset number of request visits to the service instance, that is, forward a small number of request visits to the service instance to test whether the service instance is normal and can respond to service requests in a timely manner . The preset number is 80 or 50, etc., which are set according to the performance of the service instance.

Step S20, when the service instance correctly responds to the forwarded request, resume forwarding the service request to the service instance; when the service instance correctly responds to the forwarded request, normally forward the service request to the service instance, and restore the use of the service instance described above.

Specifically, referring to Figure 2, the process of judging that the service instance correctly responds to the forwarded request includes:

Step S11 , determining the number of responded requests; after forwarding the preset number of requests to the service instance, monitoring the number of correctly forwarded requests, that is, determining the number of responded requests.

Step S12, calculating the proportion of the determined quantity in the preset quantity; calculating the quantity and proportion of correctly forwarded requests.

Step S13, when the calculated ratio is greater than the preset ratio threshold, it is determined that the service instance can correctly respond to the forwarded request. The preset ratio threshold may be 80% or 60% or the like. When the ratio of the preset number of requests that can be correctly forwarded reaches the preset ratio threshold, it is determined that the service instance can correctly respond to the forwarded request. In other embodiments of the present invention, it may also be determined according to the number of correctly forwarded requests that the service instance can correctly respond to the forwarded requests. The preset number of requests is not forwarded to the service instance at one time, but is forwarded to the service instance at intervals or continuously, and the calculation period is not limited, as long as the response request meets the condition (proportion or quantity), the service is determined The instance correctly responds to forwarded requests.

Step S30, when the service instance does not respond to the forwarded request, keep stopping forwarding the service request to the service instance. When the service instance does not respond to the forwarded request, continue to stop forwarding the service request to the service instance until the test is successful, then forward the request to the service instance normally, after a certain period of time (5 minutes or 10 minutes, etc.) , indicating that the service instance is faulty and needs to be repaired. In this embodiment, by testing the disconnected service instance, after the service instance correctly responds to the forwarded request, the service request is quickly resumed and forwarded to the service instance, making full use of resources and improving service experience.

Further, referring to Fig. 3, the method of the present invention also includes:

Step S40, obtaining the link information of the service instance within a detection period;

In this embodiment, a detection cycle is preset, and the detection cycle can be 50s or 60s. In order to save system resources, the detection cycle can be 3 minutes or 5 minutes. In order to improve the detection frequency and find problems in time, the The detection period can be set to 20s or 30s. The client initiates a business request to the server to obtain the link information of the service instance within a test period. The link information includes link success information and link failure information. The link failure information includes link timeout and link errors. When detection is triggered every time, link information of service instance in a detection cycle is obtained. A time interval can be set according to performance requirements and business requirements during each detection cycle, for example, 5 minutes or 8 minutes. A period of detection is performed at this time interval.

Step S50, judging whether the link information satisfies a preset disconnection condition;

After the link information within a detection period is acquired, it is judged whether the link information satisfies a preset disconnection condition. Specifically, in a preferred embodiment of the present invention, referring to FIG. 4, judging whether the link information satisfies a preset disconnection condition includes:

Step S51, extracting the number of link error requests from the connection information; the link information includes but not limited to link error information, link timeout, link success and other information. After obtaining the link information, extract the number of requests with link errors from the connection information, that is, extract how many requests have link errors, compare the number of link errors with a preset first number threshold, The preset first quantity threshold is set in advance, for example, 20 or 200, and is set according to the quantity of service requests initiated within the detection period, for example, if the number of requests is 200, then the first quantity threshold is set to 40 or 50, for example, if the requested quantity is 500, then the first quantity threshold is set to 100 or 120.

Step S52, when the number of link errors is greater than a preset first number threshold, it is determined that the link information satisfies a preset disconnection condition. When the number of link errors within the detection period is greater than a preset first number threshold, it is determined whether the link information satisfies a preset disconnection condition. In other embodiments of the present invention, it is also possible to calculate the proportion of link errors, and when the proportion of link errors reaches a preset proportion value (20% or 25%, etc.), it is determined that the link information meets the preset disconnection condition.

In a preferred embodiment of the present invention, referring to FIG. 5, it is also possible to determine whether the link information meets the preset disconnection condition:

Step S53, extracting the number of link timeout requests from the link information; after obtaining the link information, extracting the number of link timeout requests from the connection information, that is, extracting how many requests have link timeouts, and The number of link timeouts is compared with a preset second number threshold, and the preset second number threshold is set in advance, for example, 30 or 300, set according to the number of service requests initiated within the detection period For example, if the number of requests is 200, then the second quantity threshold is set to 60 or 70; for example, if the number of requests is 400, then the second quantity threshold is set to 120 or 140.

Step S54, when the number of link timeouts is greater than a preset second number threshold, it is determined that the link information satisfies a preset disconnection condition. When the number of link timeouts within the detection period is greater than a preset second number threshold, it is determined that the link information satisfies a preset disconnection condition. In other embodiments of the present invention, it is also possible to calculate the proportion of link errors, and when the proportion of link errors reaches a preset proportion value (20% or 25%, etc.), it is determined that the link information meets the preset disconnection condition.

In other embodiments of the present invention, the total number of link errors and link timeouts may be extracted from the link information, and when the total number is greater than a threshold (the total request is 100, the threshold is 60; or the total request is 300 , the threshold is 200, which is set according to the requirements and performance of the service instance), it is determined that the link information meets the preset disconnection condition. Likewise, the ratio can also be considered, and when the ratio exceeds 25% or 30% or even 20%, it is determined that the link information satisfies the preset disconnection condition. In the disconnection judgment process, the algorithm adopted in this embodiment is that the time point (Ticker) corresponding to each request is set to 1, and if the counter counts in the calculation unit time, as shown in Figure 6, if you want to calculate the ticker for a certain period of time For counting, you only need the ticker of the current time and the previous time, remove the data that is not in the range of this ticker, and do a count. This kind of calculation does not have any locks, and the performance is very high. For example, it is used in the calculation and counter of link information.

Step S60, stop forwarding the service request to the service instance after the link information satisfies the preset disconnection condition.

After the link information satisfies the preset disconnection condition, stop forwarding the service request to the service instance, that is, disconnect the service instance, and the service instance no longer receives service forwarding. In this embodiment, by monitoring the link information of the service instance, it is judged whether the preset disconnection condition is satisfied according to the link information, and when it is satisfied, the service instance is disconnected, and the forwarding of the service request to the service instance is stopped. By disconnecting the service instance in time, avoid the problem that the current connection between the client and the server cannot be solved in time due to the slow response of the service instance, resulting in untimely response to business requests, resulting in poor business experience, making timely response to business requests, and improving business experience. .

In order to ensure a timely response to service requests and avoid waste of resources, in a preferred embodiment of the present invention, referring to FIG. 7, the method further includes:

Step S70, determine the number of request forwardings triggered by the service instance within a detection cycle; when entering the detection cycle, determine the number of requests triggered by the service instance within this detection cycle, the triggered requests include failed and successful requests , that is, all requests forwarded to the service instance.

Step S80, when the triggered number of forwarding times is less than the preset forwarding times threshold, enter the next detection cycle to request the determination of the forwarding times; the preset forwarding times threshold is set according to the service instance performance and user requirements, for example, it can be 200 or 2000 pieces etc. After determining the triggered forwarding times of the request, it is compared that the triggered forwarding times are smaller than the preset forwarding times threshold.

Step S40, when the triggered forwarding times is greater than or equal to the preset forwarding times threshold, obtain the link information of the service instance within one detection period;

Step S50, judging whether the link information satisfies a preset disconnection condition;

Step S60, stop forwarding the service request to the service instance after the link information satisfies the preset disconnection condition. When the triggered forwarding times are greater than or equal to the preset forwarding times threshold, the connection information of the service instance in a detection cycle is obtained; when the triggered forwarding times are less than the preset forwarding times threshold, the service instance is not obtained Link information during detection cycle. In one embodiment of the present invention, a time is preset, for example, 1 minute or 30s, etc., and within this time, it is judged whether the number of requests triggering the service instance reaches a certain value, for example, 200 or 300. , enter the detection cycle, and obtain the connection information of the service instance in a detection cycle. Preferably, the detected service category is set, and only the link information of the service category is monitored and recorded, while other categories other than the set service category are not included in the link information within the detection period.

In order to better describe the embodiment of the present invention, refer to FIG. 8, which is the client and server service request architecture. During the detection period, if the connection error of the service instance or the timeout of the service instance reaches a predetermined threshold, the OSP Proxy (OSP Proxy) will trigger the fusing mechanism, and subsequent OSP Proxy will not forward requests to the service instance within a certain period of time. After the service fusing, OSP Proxy will allow a small amount of requests to access the fusing service after a certain recovery period. . If the service instance responds correctly again, the OSP Proxy resumes the normal access of the service. If the service instance still cannot respond correctly, the OSP Proxy will not forward the request to the service instance within a specific time interval. The fuse mechanism is controlled on the access interface of the service; the fuse mechanism is disabled by default, and if it needs to be used, you must configure relevant parameters in the configuration center. These parameters can be changed dynamically, and OSP Proxy does not need to be restarted; the configuration parameters of the configuration center are based on Service level, shared by multiple service instances, shared by multiple interfaces; OSP Proxy needs to report the fusing status of the service instance, and Mercury routing can know the fusing status of the relevant OSP Proxy when viewing the service

If the error rate of the instance service instance selected from Available Instances reaches the fuse policy, then move the Instance to the CircuitBreak Instances fuse instance; the system backend has a Scheduler scheduler for each CircuitBreak Instance to trigger it to be moved to In the Half-Open Instance half-disconnected instance; if the Instance in the Half-Open state is successfully called, it will be placed back in the Available Instance. During the processing of Service Instances in the Half-Open state, concurrency issues need to be considered. For example, when calling concurrently, one call succeeds and one call fails. Specifically, referring to Figure 9, when the instance3 service instance 3 of the ServiceB server B has a large error rate and meets the need for fusing, then instance3 will be fusing without affecting the invocation of the Client. After Instance3 is fused, because ServiceB only has Instance1 serving Instance 1 and Instance2 serving Instance 2, the pressure will increase. At this time, an alarm needs to be issued to avoid an avalanche effect; if the Docker application engine is introduced in the future, it can be dynamically Add Instace4 service instance 4 to relieve the pressure, and isolate Instance3 to check the problem.

The present invention further provides a server connection control device.

Referring to FIG. 10 , FIG. 10 is a schematic diagram of functional modules of a first embodiment of a connection control device for a server according to the present invention.

In an embodiment, the connection control device of the server includes: a control module 10 , a recovery module 20 , a determination module 30 , a calculation module 40 and a judgment module 50 .

The control module 10 is configured to forward a preset number of request visits to the service instance after stopping forwarding the service request to the service instance; The disconnection status is monitored in real time, so that the service instance can be used for business services as soon as possible. The set time can be 2 minutes or 3 minutes, etc. After the interval is set, forward a preset number of request visits to the service instance, that is, forward a small number of request visits to the service instance to test whether the service instance is normal and can respond to service requests in a timely manner . The preset number is 80 or 50, etc., which are set according to the performance of the service instance.

The restoration module 20 is configured to restore and forward the service request to the service instance when the service instance correctly responds to the forwarded request; and normally forward the service request to the service instance when the service instance correctly responds to the forwarded request A service request to resume usage of the service instance.

The determining module 30 is configured to determine the number of requests to respond to; after forwarding the preset number of requests to the service instance, monitor the number of correctly forwarded requests, that is, to determine the number of requests to respond to.

The calculation module 40 is used to calculate the proportion of the determined quantity in the preset quantity; calculate the quantity and proportion of correctly forwarded requests.

The judging module 50 is configured to judge that the service instance can correctly respond to the forwarded request when the calculated ratio is greater than a preset ratio threshold. The preset ratio threshold may be 80% or 60% or the like. When the ratio of the preset number of requests that can be correctly forwarded reaches the preset ratio threshold, it is determined that the service instance can correctly respond to the forwarded request. In other embodiments of the present invention, it may also be determined according to the number of correctly forwarded requests that the service instance can correctly respond to the forwarded requests. The preset number of requests is not forwarded to the service instance at one time, but is forwarded to the service instance at intervals or continuously, and the calculation period is not limited, as long as the response request meets the condition (proportion or quantity), the service is determined The instance correctly responds to forwarded requests.

The control module 10 is further configured to stop forwarding the service request to the service instance when the service instance does not respond to the forwarded request. When the service instance does not respond to the forwarded request, continue to stop forwarding the service request to the service instance until the test is successful, then forward the request to the service instance normally, after a certain period of time (5 minutes or 10 minutes, etc.) , indicating that the service instance is faulty and needs to be repaired. In this embodiment, by testing the disconnected service instance, after the service instance correctly responds to the forwarded request, the service request is quickly resumed and forwarded to the service instance, making full use of resources and improving service experience.

Referring to FIG. 11, the device further includes: an acquisition module 60,

The acquiring module 60 is configured to acquire link information of a service instance within a detection period;

In this embodiment, a detection cycle is preset, and the detection cycle can be 50s or 60s. In order to save system resources, the detection cycle can be 3 minutes or 5 minutes. In order to improve the detection frequency and find problems in time, the The detection period can be set to 20s or 30s. The client initiates a business request to the server to obtain the link information of the service instance within a test period. The link information includes link success information and link failure information. The link failure information includes link timeout and link errors. When detection is triggered every time, link information of service instance in a detection cycle is obtained. A time interval can be set according to performance requirements and business requirements during each detection cycle, for example, 5 minutes or 8 minutes. A period of detection is performed at this time interval.

The judging module 50 is configured to judge whether the link information satisfies a preset disconnection condition;

After the link information within a detection period is acquired, it is judged whether the link information satisfies a preset disconnection condition. Specifically, in a preferred embodiment of the present invention, referring to FIG. 12 , the judging module 50 includes: a judging unit 51 and a judging unit 52,

The judging unit 51 is configured to extract the number of link error requests from the connection information; the link information includes but not limited to link error information, link timeout, link success and other information. After obtaining the link information, extract the number of requests with link errors from the connection information, that is, extract how many requests have link errors, compare the number of link errors with a preset first number threshold, The preset first quantity threshold is set in advance, for example, 20 or 200, and is set according to the quantity of service requests initiated within the detection period, for example, if the number of requests is 200, then the first quantity threshold is set to 40 or 50, for example, if the requested quantity is 500, then the first quantity threshold is set to 100 or 120.

The determining unit 52 is configured to determine that the link information satisfies a preset disconnection condition when the number of link errors is greater than a preset first number threshold. When the number of link errors within the detection period is greater than a preset first number threshold, it is determined whether the link information satisfies a preset disconnection condition. In other embodiments of the present invention, it is also possible to calculate the proportion of link errors, and when the proportion of link errors reaches a preset proportion value (20% or 25%, etc.), it is determined that the link information meets the preset disconnection condition.

Further, the judging unit 51 is further configured to extract the number of link timeout requests from the link information; after obtaining the link information, extract the number of link timeout requests from the connection information, that is, extract How many requests have link timeouts, and compare the number of link timeouts with a preset second number threshold, the preset second number threshold is set in advance, for example, 30 or 300, according to the Set the number of business requests initiated during the detection period. For example, if the number of requests is 200, the second number threshold is set to 60 or 70. For example, if the number of requests is 400, the second number threshold is set to 120 or 140. .

The determining unit 52 is further configured to determine that the link information satisfies a preset disconnection condition when the number of link timeouts is greater than a preset second number threshold. When the number of link timeouts within the detection period is greater than a preset second number threshold, it is determined that the link information satisfies a preset disconnection condition. In other embodiments of the present invention, it is also possible to calculate the proportion of link errors, and when the proportion of link errors reaches a preset proportion value (20% or 25%, etc.), it is determined that the link information meets the preset disconnection condition.

In other embodiments of the present invention, the total number of link errors and link timeouts may be extracted from the link information, and when the total number is greater than a threshold (the total request is 100, the threshold is 60; or the total request is 300 , the threshold is 200, which is set according to the requirements and performance of the service instance), it is determined that the link information meets the preset disconnection condition. Likewise, the ratio can also be considered, and when the ratio exceeds 25% or 30% or even 20%, it is determined that the link information satisfies the preset disconnection condition. In the disconnection judgment process, the algorithm adopted in this embodiment is that the time point (Ticker) corresponding to each request is set to 1, and if the counter counts in the calculation unit time, as shown in Figure 6, if you want to calculate the ticker for a certain period of time For counting, you only need the ticker of the current time and the previous time, remove the data that is not in the range of this ticker, and do a count. This kind of calculation does not have any locks, and the performance is very high. For example, it is used in the calculation and counter of link information.

The control module 20 is further configured to stop forwarding the service request to the service instance after the link information satisfies a preset disconnection condition.

After the link information satisfies the preset disconnection condition, stop forwarding the service request to the service instance, that is, disconnect the service instance, and the service instance no longer receives service forwarding. In this embodiment, by monitoring the link information of the service instance, it is judged whether the preset disconnection condition is satisfied according to the link information, and when it is satisfied, the service instance is disconnected, and the forwarding of the service request to the service instance is stopped. By disconnecting the service instance in time, avoid the problem that the current connection between the client and the server cannot be solved in time due to the slow response of the service instance, resulting in untimely response to business requests, resulting in poor business experience, making timely response to business requests, and improving business experience. .

In order to ensure timely response to service requests and avoid waste of resources, in a preferred embodiment of the present invention,

The determination module 30 is also used to determine the number of request forwardings triggered by the service instance within a detection cycle; when entering the detection cycle, determine the number of requests triggered by the service instance within this detection cycle, the triggered request Includes failed and successful requests, ie all requests forwarded to the service instance.

The determining module 30 is further configured to enter the next detection cycle to request the determination of the forwarding times when the triggered forwarding times is less than the preset forwarding times threshold; the preset forwarding times threshold is set according to service instance performance and user requirements, For example, it may be 200 or 2000. After determining the triggered forwarding times of the request, it is compared that the triggered forwarding times are smaller than the preset forwarding times threshold.

The obtaining module 60 obtains the link information of the service instance within a detection period when the triggered forwarding times is greater than or equal to the preset forwarding times threshold; the judging module 50 judges whether the link information meets the preset disconnection conditions; the control module 20 After the link information satisfies a preset disconnection condition, stop forwarding the service request to the service instance. When the triggered forwarding times are greater than or equal to the preset forwarding times threshold, the connection information of the service instance in a detection cycle is obtained; when the triggered forwarding times are less than the preset forwarding times threshold, the service instance is not obtained Link information during detection cycle. In one embodiment of the present invention, a time is preset, for example, 1 minute or 30s, etc., and within this time, it is judged whether the number of requests triggering the service instance reaches a certain value, for example, 200 or 300. , enter the detection cycle, and obtain the connection information of the service instance in a detection cycle. Preferably, the detected service category is set, and only the link information of the service category is monitored and recorded, while other categories other than the set service category are not included in the link information within the detection period.

In order to better describe the embodiment of the present invention, refer to FIG. 8, which is the client and server service request architecture. During the detection period, if the connection error of the service instance or the timeout of the service instance reaches a predetermined threshold, the OSP Proxy (OSP Proxy) will trigger the fusing mechanism, and subsequent OSP Proxy will not forward requests to the service instance within a certain period of time. After the service fusing, OSP Proxy will allow a small amount of requests to access the fusing service after a certain recovery period. . If the service instance responds correctly again, the OSP Proxy resumes the normal access of the service. If the service instance still cannot respond correctly, the OSP Proxy will not forward the request to the service instance within a specific time interval. The fuse mechanism is controlled on the access interface of the service; the fuse mechanism is disabled by default, and if it needs to be used, you must configure relevant parameters in the configuration center. These parameters can be changed dynamically, and OSP Proxy does not need to be restarted; the configuration parameters of the configuration center are based on Service level, shared by multiple service instances, shared by multiple interfaces; OSP Proxy needs to report the fusing status of the service instance, and Mercury routing can know the fusing status of the relevant OSP Proxy when viewing the service

If the error rate of the instance service instance selected from Available Instances reaches the fuse policy, then move the Instance to the CircuitBreak Instances fuse instance; the system backend has a Scheduler scheduler for each CircuitBreak Instance to trigger it to be moved to In the Half-Open Instance half-disconnected instance; if the Instance in the Half-Open state is successfully called, it will be placed back in the Available Instance. During the processing of Service Instances in the Half-Open state, concurrency issues need to be considered. For example, when calling concurrently, one call succeeds and one call fails. Specifically, referring to Figure 9, when the instance3 service instance 3 of the ServiceB server B has a large error rate and meets the need for fusing, then instance3 will be fusing without affecting the invocation of the Client. After Instance3 is fused, because ServiceB only has Instance1 serving Instance 1 and Instance2 serving Instance 2, the pressure will increase. At this time, an alarm needs to be issued to avoid an avalanche effect; if the Docker application engine is introduced in the future, it can be dynamically Add Instace4 service instance 4 to relieve the pressure, and isolate Instance3 to check the problem.

The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process conversion made by using the description of the present invention and the contents of the accompanying drawings, or directly or indirectly used in other related technical fields , are all included in the scope of patent protection of the present invention in the same way.

Claims (10)

1. a kind of connection control method of server is it is characterised in that include step：

Stop forwarding service request to Service Instance after, the request of predetermined number is accessed and is forwarded on described Service Instance；

When described Service Instance correctly responds the request of forwarding, recover to forward service request to described Service Instance；

When described Service Instance does not respond the request of forwarding, keep stopping forwarding service request to described Service Instance.

2. the connection control method of server as claimed in claim 1 is it is characterised in that the described request by predetermined number is visited Ask after being forwarded on described Service Instance, also include：

Determine the quantity of the request of response；

Calculate the shared ratio in predetermined number of the quantity determining；

When the ratio calculating is more than preset ratio threshold value, judge that described Service Instance can correctly respond the request of forwarding.

3. the connection control method of server as claimed in claim 1 or 2 is it is characterised in that methods described also includes：

Obtain link information in a detection cycle for the Service Instance；

Judge whether described link information meets default disconnection condition；

After described link information meets default disconnection condition, stop forwarding service request to described Service Instance.

4. the connection control method of server as claimed in claim 3 is it is characterised in that the described link information of described judgement is No meet default disconnection condition and include：

The quantity that break links request occurs is extracted from described link information；

When the quantity of described break links is more than default first amount threshold, judge whether described link information meets default Disconnection condition.

5. the connection control method of server as claimed in claim 3 is it is characterised in that the described link information of described judgement is No meet default disconnection condition and include：

The quantity that link TIMEOUT occurs is extracted from described link information；

When the quantity of described link time-out is more than default second amount threshold, judge whether described link information meets default Disconnection condition.

6. a kind of connection control device of server is it is characterised in that include：

Control module, for stop forwarding service request to Service Instance after, the request of predetermined number is accessed and is forwarded to institute State on Service Instance；

Recovery module, for correctly respond in described Service Instance forwarding request when, recover to forward service request to described clothes In pragmatic example；

Described control module, is additionally operable to, when described Service Instance does not respond the request of forwarding, keep stopping forwarding service request To described Service Instance.

7. the connection control device of server as claimed in claim 6 is it is characterised in that described device also includes：

Determining module, for determining the quantity of the request of response；

Computing module, the shared ratio in predetermined number of the quantity for calculating determination；

Judge module, turns for when the ratio calculating is more than preset ratio threshold value, judging that described Service Instance can correctly respond The request sent out.

8. the connection control device of server as claimed in claims 6 or 7 is it is characterised in that device also includes：

Acquisition module, for obtaining link information in a detection cycle for the Service Instance；

Judge module, for judging whether described link information meets default disconnection condition；

Described control module, is additionally operable to, after described link information meets default disconnection condition, stop forwarding service request extremely On described Service Instance.

9. the connection control device of server as claimed in claim 8 is it is characterised in that described judge module, including：

Judging unit, for extracting the quantity that break links request occurs from described link information；

Identifying unit, for when the quantity of described break links is more than default first amount threshold, judging described link letter Whether breath meets default disconnection condition.

10. the connection control device of server as claimed in claim 9 is it is characterised in that described judging unit, be additionally operable to from The quantity that link TIMEOUT occurs is extracted in described link information；

Described identifying unit, is additionally operable to, when the quantity of described link time-out is more than default second amount threshold, judge described Whether link information meets default disconnection condition.