Use and expansion of fault tolerance and fusing

@[toc]

One 、 Introduction of fault tolerance mechanism

• Concept when the client invokes a microservice , There's a problem , Be able to fail over , Fault tolerance mechanism . Pictured ：
  

• Aim to ensure high availability of microservices .

Two 、 Why use fault tolerance

• The failover mechanism is shown in the figure ：

• 

3、 ... and 、 How to apply fault tolerance mechanism in microservices

• Implementation conditions

• while

• Implementation steps

 int count = 0;
 // Fail over
 for(int i=0;i<=3;i++)
 {
 // Judge whether the threshold is reached
 if (count == 3)
 {
 // Exit loop , Return exception information
 throw Exception(&quot; The microservice retry operation exceeded the threshold &quot;);
 } 
 // Business code
 try
 {
 // HttpClient  Request microservice code
 ...................
 }
 catch(Exception ex)
 {
 // Capture exception information  
 count ++; 
 }
 } 


Four 、 Defects of fault-tolerant mechanism

• defects

• Low performance , Resource consumption

• Solution

• Set a fixed number of times

• Use circuit breakers

5、 ... and 、 How to use in micro service system Polly

• Fuse

• The concept of fusing is a request . When the request arrives at the server , The request was retried 3 Time 【 Number of tests 】, No request succeeded , Directly fuse the request , Please come in next time , Don't try again , Return exception information directly .

• Conditions

• Polly

• Code implementation

• Conditions

• install  Polly

• step

• install

 Microsoft.Extensions.Http.Polly

-  stay Startup.cs  Register in the file
 -  Method name  ConfigureServices

 // Abnormal degradation information  
 var fallBack = new HttpResponseMessage(){ 
 Content = new StringContent(" The system is busy ！"),
 StatusCode = 504
 };
 //5： Circuit breaker threshold
 //10： Melting time
 //ExecutionRejectedException: Capture all abnormal information of fuse
 //10： Total number of threads used
 services.AddHttpClient("micro[ Request reuse ： Custom name ]")
 .AddPolicyHandler(Policy<HttpResponseMeaage>.Handle<ExecutionRejectedException>().Fallback(fallBack)) // Abnormal information degradation
 .AddPolicyHandler(Policy<HttpResponseMeaage>.Handle<Exception>().CircuitBreakerAsync(5,TimeSpan.FromSeconds(10))) // Circuit breaker
 .AddPolicyHandler(Policy.TimeoutAsync<HttpResponseMeaage>(60))// Set timeout
 .AddPolicyHandler(Policy<HttpResponseMeaage>.Handle<Exception>().RetryAsync(1))// Retry count
 .AddPolicyHandler(Policy.BulkheadAsync<HttpResponseMeaage>(10,100));// resources （ Threads ） Isolation  //10： Total number of threads used  100: Total request cache

 -  Business code

 private HttpClientFactory _httpClientFactory;
 // Inject... Into the constructor
  Constructors (HttpClientFactory httpClientFactory)
 {
 _httpClientFactory = httpClientFactory;
 } 
 
 int count = 0;
 // Fail over
 for(int i=0;i<=3;i++)
 {
 // Judge whether the threshold is reached
 if (count == 3)
 {
 // Exit loop , Return exception information
 throw Exception(&quot; The microservice retry operation exceeded the threshold &quot;);
 } 
 // Business code  
 try
 {
 // Microservice request
 HttpClient client = _httpClientFactory.CreateClient(&quot;micro&quot;);
 .........
 }
 catch(HttpRequestException ex)
 { 
 // Capture exception information  
 count ++; 
 }
 } 


6、 ... and 、 Microservice fault tolerance mechanism Polly Expand

• If two microservices are called at the same time in the aggregated microservice , When the first service hangs up , In order not to affect the subsequent service calls , We need to do service isolation , The code is as follows ：

• stay Startup.cs  Register in the file

• Method name  ConfigureServices

 services.AddHttpClient(" Define different names according to different services ")
 .AddPolicyHandler(Policy<HttpResponseMeaage>.Handle<ExecutionRejectedException>().Fallback(fallBack)) // Abnormal information degradation
 .AddPolicyHandler(Policy<HttpResponseMeaage>.Handle<Exception>().CircuitBreakerAsync(5,TimeSpan.FromSeconds(10))) // Circuit breaker
 .AddPolicyHandler(Policy.TimeoutAsync<HttpResponseMeaage>(60))// Set timeout
 .AddPolicyHandler(Policy<HttpResponseMeaage>.Handle<Exception>().RetryAsync(1))// Retry count
 .AddPolicyHandler(Policy.Bulkhead<HttpResponseMeaage>(10,100));// Resource isolation

 == Be careful ： When http On request , The name of each service requested must be consistent with the registered name .==


• Caller code

 HttpClient client = _httpClientFactory.CreateClient(" Define different names according to different services ");