详解feign调用session丢失解决方案

时间:2022-01-13 16:01:39

最近在做项目的时候发现,微服务使用feign相互之间调用时,存在session丢失的问题。例如,使用feign调用某个远程api,这个远程api需要传递一个鉴权信息,我们可以把cookie里面的session信息放到header里面,这个header是动态的,跟你的httprequest相关,我们选择编写一个拦截器来实现header的传递,也就是需要实现requestinterceptor接口,具体代码如下:

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
@configuration
@enablefeignclients(basepackages = "com.xxx.xxx.client")
public class feignclientsconfigurationcustom implements requestinterceptor {
 
 @override
 public void apply(requesttemplate template) {
 
  requestattributes requestattributes = requestcontextholder.getrequestattributes();
  if (requestattributes == null) {
   return;
  }
 
  httpservletrequest request = ((servletrequestattributes) requestattributes).getrequest();
  enumeration<string> headernames = request.getheadernames();
  if (headernames != null) {
   while (headernames.hasmoreelements()) {
    string name = headernames.nextelement();
    enumeration<string> values = request.getheaders(name);
    while (values.hasmoreelements()) {
     string value = values.nextelement();
     template.header(name, value);
    }
   }
  }
 
 }
 
}

经过测试,上面的解决方案可以正常的使用; 

但是,当引入hystrix熔断策略时,出现了一个新的问题:

?
1
requestattributes requestattributes = requestcontextholder.getrequestattributes();

此时requestattributes会返回null,从而无法传递session信息,最终发现requestcontextholder.getrequestattributes(),该方法是从threadlocal变量里面取得对应信息的,这就找到问题原因了,是由于hystrix熔断机制导致的。 
hystrix有2个隔离策略:thread以及semaphore,当隔离策略为 thread 时,是没办法拿到 threadlocal 中的值的。

因此有两种解决方案:

方案一:调整格隔离策略:

?
1
hystrix.command.default.execution.isolation.strategy: semaphore

这样配置后,feign可以正常工作。

但该方案不是特别好。原因是hystrix官方强烈建议使用thread作为隔离策略! 可以参考官方文档说明。

方案二:自定义策略

记得之前在研究zipkin日志追踪的时候,看到过sleuth有自己的熔断机制,用来在thread之间传递trace信息,sleuth是可以拿到自己上下文信息的,查看源码找到了 
org.springframework.cloud.sleuth.instrument.hystrix.sleuthhystrixconcurrencystrategy 
这个类,查看sleuthhystrixconcurrencystrategy的源码,继承了hystrixconcurrencystrategy,用来实现了自己的并发策略。

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
/**
 * abstract class for defining different behavior or implementations for concurrency related aspects of the system with default implementations.
 * <p>
 * for example, every {@link callable} executed by {@link hystrixcommand} will call {@link #wrapcallable(callable)} to give a chance for custom implementations to decorate the {@link callable} with
 * additional behavior.
 * <p>
 * when you implement a concrete {@link hystrixconcurrencystrategy}, you should make the strategy idempotent w.r.t threadlocals.
 * since the usage of threads by hystrix is internal, hystrix does not attempt to apply the strategy in an idempotent way.
 * instead, you should write your strategy to work idempotently. see https://github.com/netflix/hystrix/issues/351 for a more detailed discussion.
 * <p>
 * see {@link hystrixplugins} or the hystrix github wiki for information on configuring plugins: <a
 * href="https://github.com/netflix/hystrix/wiki/plugins" rel="external nofollow" >https://github.com/netflix/hystrix/wiki/plugins</a>.
 */
public abstract class hystrixconcurrencystrategy

搜索发现有好几个地方继承了hystrixconcurrencystrategy类 

详解feign调用session丢失解决方案

其中就有我们熟悉的spring security和刚才提到的sleuth都是使用了自定义的策略,同时由于hystrix只允许有一个并发策略,因此为了不影响spring security和sleuth,我们可以参考他们的策略实现自己的策略,大致思路: 

将现有的并发策略作为新并发策略的成员变量; 

在新并发策略中,返回现有并发策略的线程池、queue; 

代码如下:

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
public class feignhystrixconcurrencystrategy extends hystrixconcurrencystrategy {
 
 private static final logger log = loggerfactory.getlogger(feignhystrixconcurrencystrategy.class);
 private hystrixconcurrencystrategy delegate;
 
 public feignhystrixconcurrencystrategy() {
  try {
   this.delegate = hystrixplugins.getinstance().getconcurrencystrategy();
   if (this.delegate instanceof feignhystrixconcurrencystrategy) {
    // welcome to singleton hell...
    return;
   }
   hystrixcommandexecutionhook commandexecutionhook =
     hystrixplugins.getinstance().getcommandexecutionhook();
   hystrixeventnotifier eventnotifier = hystrixplugins.getinstance().geteventnotifier();
   hystrixmetricspublisher metricspublisher = hystrixplugins.getinstance().getmetricspublisher();
   hystrixpropertiesstrategy propertiesstrategy =
     hystrixplugins.getinstance().getpropertiesstrategy();
   this.logcurrentstateofhystrixplugins(eventnotifier, metricspublisher, propertiesstrategy);
   hystrixplugins.reset();
   hystrixplugins.getinstance().registerconcurrencystrategy(this);
   hystrixplugins.getinstance().registercommandexecutionhook(commandexecutionhook);
   hystrixplugins.getinstance().registereventnotifier(eventnotifier);
   hystrixplugins.getinstance().registermetricspublisher(metricspublisher);
   hystrixplugins.getinstance().registerpropertiesstrategy(propertiesstrategy);
  } catch (exception e) {
   log.error("failed to register sleuth hystrix concurrency strategy", e);
  }
 }
 
 private void logcurrentstateofhystrixplugins(hystrixeventnotifier eventnotifier,
   hystrixmetricspublisher metricspublisher, hystrixpropertiesstrategy propertiesstrategy) {
  if (log.isdebugenabled()) {
   log.debug("current hystrix plugins configuration is [" + "concurrencystrategy ["
     + this.delegate + "]," + "eventnotifier [" + eventnotifier + "]," + "metricpublisher ["
     + metricspublisher + "]," + "propertiesstrategy [" + propertiesstrategy + "]," + "]");
   log.debug("registering sleuth hystrix concurrency strategy.");
  }
 }
 
 @override
 public <t> callable<t> wrapcallable(callable<t> callable) {
  requestattributes requestattributes = requestcontextholder.getrequestattributes();
  return new wrappedcallable<>(callable, requestattributes);
 }
 
 @override
 public threadpoolexecutor getthreadpool(hystrixthreadpoolkey threadpoolkey,
   hystrixproperty<integer> corepoolsize, hystrixproperty<integer> maximumpoolsize,
   hystrixproperty<integer> keepalivetime, timeunit unit, blockingqueue<runnable> workqueue) {
  return this.delegate.getthreadpool(threadpoolkey, corepoolsize, maximumpoolsize, keepalivetime,
    unit, workqueue);
 }
 
 @override
 public threadpoolexecutor getthreadpool(hystrixthreadpoolkey threadpoolkey,
   hystrixthreadpoolproperties threadpoolproperties) {
  return this.delegate.getthreadpool(threadpoolkey, threadpoolproperties);
 }
 
 @override
 public blockingqueue<runnable> getblockingqueue(int maxqueuesize) {
  return this.delegate.getblockingqueue(maxqueuesize);
 }
 
 @override
 public <t> hystrixrequestvariable<t> getrequestvariable(hystrixrequestvariablelifecycle<t> rv) {
  return this.delegate.getrequestvariable(rv);
 }
 
 static class wrappedcallable<t> implements callable<t> {
  private final callable<t> target;
  private final requestattributes requestattributes;
 
  public wrappedcallable(callable<t> target, requestattributes requestattributes) {
   this.target = target;
   this.requestattributes = requestattributes;
  }
 
  @override
  public t call() throws exception {
   try {
    requestcontextholder.setrequestattributes(requestattributes);
    return target.call();
   } finally {
    requestcontextholder.resetrequestattributes();
   }
  }
 }
}

最后,将这个策略类作为bean配置到feign的配置类feignclientsconfigurationcustom中

?
1
2
3
4
@bean
public feignhystrixconcurrencystrategy feignhystrixconcurrencystrategy() {
 return new feignhystrixconcurrencystrategy();
}

至此,结合feignclientsconfigurationcustom配置feign调用session丢失的问题完美解决。

以上就是本文的全部内容,希望对大家的学习有所帮助,也希望大家多多支持服务器之家。 

原文链接:https://blog.csdn.net/zl1zl2zl3/article/details/79084368