前言
之前的几篇文章分析了spring boot 中有关endpoint的实现,细心的朋友可以发现,在org.springframework.boot.actuate.endpoint.mvc 包下也有一系列的xxxEndpoint,这又是为什么呢?
原因是: 我们很多情况下,都是访问接口的方式获取应用的监控,之前的分析是其实现的底层,要想实现通过接口访问,还需要对其进行包装一番,org.springframework.boot.actuate.endpoint.mvc 包下的实现就是干的这种事,下面,我们就来分析一下吧
解析
关于mvcEndPoint的类图如下,
下面我们就来1个1个的来分析吧.
MvcEndpoint
MvcEndpoint –> 顶层接口,实现类允许使用@RequestMapping和完整的Spring MVC机制,但不能在类型级别使用@Controller或@RequestMapping,因为这将导致路径的双重映射,一次通过常规MVC处理程序映射,一次通过EndpointHandlerMapping。
该类声明如下:
public interface MvcEndpoint { // 禁用端点的响应实体 ResponseEntity<Map<String, String>> DISABLED_RESPONSE = new ResponseEntity<Map<String, String>>( Collections.singletonMap("message", "This endpoint is disabled"), HttpStatus.NOT_FOUND); // 返回端点的MVC路径 String getPath(); // 返回端点是否暴露敏感信息 boolean isSensitive(); // 返回端点暴露的类型。或者返回null 如果当前的MvcEndpoint 暴露的信息 不能视为一个传统的Endpoint(也就是我们之前分析的那一堆Endpoint) Class<? extends Endpoint> getEndpointType(); }
AbstractMvcEndpoint
AbstractMvcEndpoint继承了WebMvcConfigurerAdapter,实现了MvcEndpoint, EnvironmentAware接口,此时 AbstractMvcEndpoint 就持有了Environment,可以对spring mvc 做个性化设置.
字段,构造器分别如下:
// Endpoint 请求路径 private String path; // endpoint是否可用 private Boolean enabled; // 标识该endpoint 是否暴露了敏感数据,如果为true,访问该Endpoint需要进行校验 private Boolean sensitive; // 是否默认敏感 private final boolean sensitiveDefault; public AbstractMvcEndpoint(String path, boolean sensitive) { setPath(path); this.sensitiveDefault = sensitive; } public AbstractMvcEndpoint(String path, boolean sensitive, boolean enabled) { setPath(path); this.sensitiveDefault = sensitive; this.enabled = enabled; }
AbstractMvcEndpoint 实现了MvcEndpoint的方法,分别如下:
getPath,直接返回其字段值即可,代码如下:
public String getPath() { return this.path; }
isSensitive–> 默认返回false.代码如下:
public boolean isSensitive() { // 默认返回false return EndpointProperties.isSensitive(this.environment, this.sensitive, this.sensitiveDefault); }
调用:
public static boolean isSensitive(Environment environment, Boolean sensitive, boolean sensitiveDefault) { // 1. 如果sensitive 不等于null,则直接返回 if (sensitive != null) { return sensitive; } // 2. 如果environment 不等于null 并且 environment中配置有endpoints.sensitive的属性,则 // 返回其配置值 if (environment != null && environment.containsProperty(ENDPOINTS_SENSITIVE_PROPERTY)) { return environment.getProperty(ENDPOINTS_SENSITIVE_PROPERTY, Boolean.class); } // 3. 返回给的默认值 return sensitiveDefault; }
- 如果sensitive 不等于null,则直接返回
- 如果environment 不等于null 并且 environment中配置有endpoints.sensitive的属性,则 返回其配置值
- 返回给的默认值
getEndpointType –> 默认返回null,代码如下:
public Class<? extends Endpoint> getEndpointType() { return null; }
此外,该类还声明了1个方法,判断当前端点是否可用,默认返回true代码如下:
public boolean isEnabled() { // 默认返回true return EndpointProperties.isEnabled(this.environment, this.enabled); }
调用:
public static boolean isEnabled(Environment environment, Boolean enabled) { // 1. 如果enabled 不为null,则进行返回. if (enabled != null) { return enabled; } // 2. 如果Environment 不等于null 并且Environment 配置有endpoints.enabled的属性, // 则返回其配置的值 if (environment != null && environment.containsProperty(ENDPOINTS_ENABLED_PROPERTY)) { return environment.getProperty(ENDPOINTS_ENABLED_PROPERTY, Boolean.class); } // 3. 默认为true return true; }
- 如果enabled 不为null,则进行返回.
- 如果Environment 不等于null 并且Environment 配置有endpoints.enabled的属性,则返回其配置的值
- 默认为true
NamedMvcEndpoint
NamedMvcEndpoint继承自MvcEndpoint,使一个MvcEndpoint可以包含一个逻辑名.不像getPath()–>它没有给用户一个机会去改变endpoint的名字.NamedMvcEndpoint 提供了一个一致的方式去引用一个endpoint.
该类声明了1个方法,如下:
// 返回一个逻辑名字,不能为null,空,字母数字组成 String getName();
AbstractNamedMvcEndpoint
- 该类继承自AbstractMvcEndpoint,实现了NamedMvcEndpoint接口.
字段,构造器如下:
// Endpoint 名字 private final String name; public AbstractNamedMvcEndpoint(String name, String path, boolean sensitive) { super(path, sensitive); Assert.hasLength(name, "Name must not be empty"); this.name = name; } public AbstractNamedMvcEndpoint(String name, String path, boolean sensitive, boolean enabled) { super(path, sensitive, enabled); Assert.hasLength(name, "Name must not be empty"); this.name = name; }
getName,只需返回name属性值即可,很简单,代码如下:
public String getName() { return this.name; }
该类的子类有:
- AuditEventsMvcEndpoint
- DocsMvcEndpoint
- HalJsonMvcEndpoint
- HeapdumpMvcEndpoint
- JolokiaMvcEndpoint
- LogFileMvcEndpoint
这些子类的实现我们后续进行分析…
AbstractEndpointMvcAdapter
- AbstractEndpointMvcAdapter 实现了NamedMvcEndpoint接口,是MvcEndpoint的抽象基类
字段,构造器如下:
// 被代理的底层端点(端点子类 private final E delegate; // 端点URL路径 private String path; // public AbstractEndpointMvcAdapter(E delegate) { Assert.notNull(delegate, "Delegate must not be null"); this.delegate = delegate; }
NamedMvcEndpoint接口方法实现如下:
getName,返回被代理的Endpoint的id,代码如下:
public String getName() { return this.delegate.getId(); }
getPath –> 如果path不等于null,则直接返回,否则使用/+Endpoint的id,代码如下:
public String getPath() { // 如果path不等于null,则直接返回,否则使用/+Endpoint的id return (this.path != null ? this.path : "/" + this.delegate.getId()); }
isSensitive–> 直接调用被代理的Endpoint的isSensitive方法即可.代码如下:
public boolean isSensitive() { return this.delegate.isSensitive(); }
getEndpointType –> 返回被代理的 Endpoint的类型即可,代码如下:
public Class<? extends Endpoint> getEndpointType() { return this.delegate.getClass(); }
AbstractEndpointMvcAdapter 还声明了2个方法,供子类使用
getDisabledResponse–> 返回该相应当所代理的endpoint不可用时.代码如下:
protected ResponseEntity<?> getDisabledResponse() { return MvcEndpoint.DISABLED_RESPONSE; }
即:
ResponseEntity<Map<String, String>> DISABLED_RESPONSE = new ResponseEntity<Map<String, String>>( Collections.singletonMap("message", "This endpoint is disabled"), HttpStatus.NOT_FOUND);
invoke–> 调用代理的Endpoint,并返回调用结果,代码如下:
protected Object invoke() { if (!this.delegate.isEnabled()) { return getDisabledResponse(); } return this.delegate.invoke(); }
- 端点不可用(禁用),则返回默认的不可用信息.当Endpoint被禁用时,是不会注册的.
- 否则,调用Endpoint的invoke方法
EndpointMvcAdapter
EndpointMvcAdapter,继承自AbstractEndpointMvcAdapter,将Endpoint适配为MvcEndpoint. 构造器如下:
public EndpointMvcAdapter(Endpoint<?> delegate) { super(delegate); }
该类覆写了invoke,使其能过被spring mvc 处理–> 暴露接口(关于这部分,我们后续有分析),代码如下:
@Override @ActuatorGetMapping @ResponseBody public Object invoke() { return super.invoke(); }
其中@ActuatorGetMapping 就是@RequestMapping的封装,被该注解标注的方法,其请求方式为get,产生的数据格式为application/vnd.spring-boot.actuator.v1+json和application/json.代码如下:
@Target(ElementType.METHOD) @Retention(RetentionPolicy.RUNTIME) @Documented @RequestMapping(method = RequestMethod.GET, produces = { ActuatorMediaTypes.APPLICATION_ACTUATOR_V1_JSON_VALUE, MediaType.APPLICATION_JSON_VALUE }) @interface ActuatorGetMapping { /** * Alias for {@link RequestMapping#value}. * @return the value */ @AliasFor(annotation = RequestMapping.class) String[] value() default {}; }
EnvironmentMvcEndpoint
该类继承自EndpointMvcAdapter,实现了EnvironmentAware,因此,该类也就持有了Environment.
字段,构造器如下:
private Environment environment; public EnvironmentMvcEndpoint(EnvironmentEndpoint delegate) { super(delegate); }
此时, EnvironmentMvcEndpoint 也就持有了EnvironmentEndpoint的实例
该类声明了1个被@ActuatorGetMapping注解的方法,value,代码如下:
@ActuatorGetMapping("/{name:.*}") @ResponseBody @HypermediaDisabled public Object value(@PathVariable String name) { if (!getDelegate().isEnabled()) { // Shouldn't happen - MVC endpoint shouldn't be registered when delegate's // disabled return getDisabledResponse(); } return new NamePatternEnvironmentFilter(this.environment).getResults(name); }
- @ActuatorGetMapping(“/{name:.*}”)与@PathVariable String name –> rest风格,将其name注入到方法的参数name中,匹配规则是任意字符
- @ResponseBody –> 返回json格式的数据
@HypermediaDisabled–>表明该MvcEndpoint或者@RequestMapping注解的方法不会生成hypermedia 的响应.代码如下:
@Target({ ElementType.TYPE, ElementType.METHOD }) @Retention(RetentionPolicy.RUNTIME) @Documented public @interface HypermediaDisabled { }
其方法逻辑如下:
- 如果EnvironmentEndpoint不可用,则返回Disabled Response.
否则实例化NamePatternEnvironmentFilter,调用其getResults方法获得对应name的属性值.
NamePatternEnvironmentFilter继承自NamePatternFilter.
NamePatternFilter:
NamePatternFilter–> 可以使用name正则表达式过滤源数据的实用工具类,用来检测检测名称是否是经典的“single value”键或正则表达式.子类必须实现getValue,getNames 方法.该类是1个泛型类,其泛型参数T 代表着原始数据的类型.
字段,构造器如下:
private static final String[] REGEX_PARTS = { "*", "$", "^", "+", "[" }; private final T source; NamePatternFilter(T source) { this.source = source; }
声明了3个抽象方法:
protected abstract void getNames(T source, NameCallback callback); protected abstract Object getValue(T source, String name); protected abstract Object getOptionalValue(T source, String name);
NamePatternFilter中声明了1个方法,代码如下:
public Map<String, Object> getResults(String name) { // 1. 如果name含有 "*", "$", "^", "+", "[" ,则认为是一个正则表达式,将其返回Pattern.否则返回null Pattern pattern = compilePatternIfNecessary(name); if (pattern == null) { // 2. 如果pattern 等于null,则说明name 是一个普通字符串,调用getValue 这一抽象方法获得value,放入result后返回 Object value = getValue(this.source, name); Map<String, Object> result = new HashMap<String, Object>(); result.put(name, value); return result; } // 3. 实例化 ResultCollectingNameCallback ResultCollectingNameCallback resultCollector = new ResultCollectingNameCallback( pattern); // 4. 抽象方法 getNames(this.source, resultCollector); // 5, 返回结果 return resultCollector.getResults(); }
如果name含有 “*”, “$”, “^”, “+”, “[” ,则认为是一个正则表达式,将其返回Pattern.否则返回null,代码如下:
private Pattern compilePatternIfNecessary(String name) { for (String part : REGEX_PARTS) { if (name.contains(part)) { try { return Pattern.compile(name); } catch (PatternSyntaxException ex) { return null; } } } return null; }
- 如果pattern 等于null,则说明name 是一个普通字符串,调用getValue 这一抽象方法获得value,放入result后返回
实例化 ResultCollectingNameCallback,该类实现了NameCallback接口,该接口只声明了如下方法:
void addName(String name);
ResultCollectingNameCallback中的字段,构造器如下:
// 将name 转换为正则所对应的对象 private final Pattern pattern; // 结果集 private final Map<String, Object> results = new LinkedHashMap<String, Object>(); ResultCollectingNameCallback(Pattern pattern) { this.pattern = pattern; }
其addName方法如下:
public void addName(String name) { // 1. 如果name 符合正则,则通过调用getOptionalValue 获得值后加入到results if (this.pattern.matcher(name).matches()) { Object value = getOptionalValue(NamePatternFilter.this.source, name); if (value != null) { this.results.put(name, value); } } }
如果name 符合正则,则通过调用getOptionalValue 获得值后加入到results
此外还声明了get方法,来暴露结果集,代码如下:
public Map<String, Object> getResults() { return this.results; }
执行getNames方法,进行结果的收集
- 返回结果
NamePatternEnvironmentFilter
- 继承了NamePatternFilter 接口,泛型参数为Environment
抽象方法分别实现如下:
getValue:
protected Object getValue(Environment source, String name) { // 1. 获取值,如果没获取到,则抛出NoSuchPropertyException,否则,对其进行脱敏 Object result = getValue(name); if (result == null) { throw new NoSuchPropertyException("No such property: " + name); } return ((EnvironmentEndpoint) getDelegate()).sanitize(name, result); }
调用getValue方法获取值,如果没获取到,则抛出NoSuchPropertyException,否则,对其进行脱敏.
getValue–>直接从environment 中获取值,代码如下:
private Object getValue(String name) { // 直接从environment 中获取值, return ((EnvironmentEndpoint) getDelegate()).getResolver().getProperty(name, Object.class); }
getNames–> 遍历source中的PropertySources,将PropertySource的属性名依次的加入到ResultCollectingNameCallback中.代码如下:
protected void getNames(Environment source, NameCallback callback) { if (source instanceof ConfigurableEnvironment) { // 遍历source中的PropertySources,将PropertySource的属性名依次的加入到ResultCollectingNameCallback中 getNames(((ConfigurableEnvironment) source).getPropertySources(), callback); } } private void getNames(PropertySources propertySources, NameCallback callback) { for (PropertySource<?> propertySource : propertySources) { if (propertySource instanceof EnumerablePropertySource) { EnumerablePropertySource<?> source = (EnumerablePropertySource<?>) propertySource; for (String name : source.getPropertyNames()) { callback.addName(name); } } } }
getOptionalValue,代码如下:
protected Object getOptionalValue(Environment source, String name) { // 1. 获得name对应的属性值 Object result = getValue(name); if (result != null) { // 2. 如果属性值存在则进行脱敏 result = ((EnvironmentEndpoint) getDelegate()).sanitize(name, result); } // 3. 否则直接返回null return result; }
- 获得name对应的属性值
- 如果属性值存在则进行脱敏
- 否则直接返回null
属性配置(有@ConfigurationProperties(prefix = “endpoints.env”)注解):
endpoints.env.path=/env
自动装配:
在EndpointWebMvcManagementContextConfiguration中进行了配置,如下:
@Bean @ConditionalOnMissingBean @ConditionalOnBean(EnvironmentEndpoint.class) @ConditionalOnEnabledEndpoint("env") public EnvironmentMvcEndpoint environmentMvcEndpoint(EnvironmentEndpoint delegate) { return new EnvironmentMvcEndpoint(delegate); }
- @Bean –> 注册1个id为 environmentMvcEndpoint,类型为EnvironmentMvcEndpoint的bean
- @ConditionalOnMissingBean–>BeanFactory中不存在EnvironmentMvcEndpoint类型的bean时生效
- @ConditionalOnBean(EnvironmentEndpoint.class) –> 当BeanFactory中存在EnvironmentEndpoint类型的Bean时生效
- @ConditionalOnEnabledEndpoint(“env”)–> 如果配置有endpoints.env.enabled = true 或者endpoints.enabled= true 则该配置生效.关于此处的实现我们后续有文章进行分析.
LoggersMvcEndpoint
字段,构造器如下:
private final LoggersEndpoint delegate; public LoggersMvcEndpoint(LoggersEndpoint delegate) { super(delegate); this.delegate = delegate; }
该类声明了2个@ActuatorGetMapping(“/{name:.*}”)注解的方法:
get,代码如下:
@ActuatorGetMapping("/{name:.*}") @ResponseBody @HypermediaDisabled public Object get(@PathVariable String name) { if (!this.delegate.isEnabled()) { // Shouldn't happen - MVC endpoint shouldn't be registered when delegate's // disabled return getDisabledResponse(); } LoggerLevels levels = this.delegate.invoke(name); return (levels == null ? ResponseEntity.notFound().build() : levels); }
- @ActuatorGetMapping(“/{name:.*}”) 与 @PathVariable String name–> rest风格,将其name注入到方法的参数name中,匹配规则是任意字符
方法逻辑如下:
- 如果LoggersEndpoint不可用,则返回默认的不可用消息
否则,调用LoggersEndpoint#invoke 获得LoggerLevels.代码如下:
public LoggerLevels invoke(String name) { Assert.notNull(name, "Name must not be null"); LoggerConfiguration configuration = this.loggingSystem .getLoggerConfiguration(name); return (configuration == null ? null : new LoggerLevels(configuration)); }
- 调用LoggingSystem#getLoggerConfiguration 获得LoggerConfiguration
- 如果LoggerConfiguration等于null,则返回null,否则,返回LoggerLevels
由于spring boot 默认使用的是logback,因此,此处调用的是LogbackLoggingSystem中的实现,代码如下:
public LoggerConfiguration getLoggerConfiguration(String loggerName) { return getLoggerConfiguration(getLogger(loggerName)); }
调用getLogger 获得对应的logger,代码如下:
private ch.qos.logback.classic.Logger getLogger(String name) { LoggerContext factory = getLoggerContext(); if (StringUtils.isEmpty(name) || ROOT_LOGGER_NAME.equals(name)) { name = Logger.ROOT_LOGGER_NAME; } return factory.getLogger(name); }
- 获得LoggerContext
- 如果name为空,或者name等于ROOT,则name赋值为ROOT
- 根据name获得对应的Logger
调用getLoggerConfiguration获得对应的LoggerConfiguration.
private LoggerConfiguration getLoggerConfiguration( ch.qos.logback.classic.Logger logger) { if (logger == null) { return null; } LogLevel level = LEVELS.convertNativeToSystem(logger.getLevel()); LogLevel effectiveLevel = LEVELS .convertNativeToSystem(logger.getEffectiveLevel()); String name = logger.getName(); if (!StringUtils.hasLength(name) || Logger.ROOT_LOGGER_NAME.equals(name)) { name = ROOT_LOGGER_NAME; } return new LoggerConfiguration(name, level, effectiveLevel); }
- 如果logger等于null,返回null
- 根据logger对应的level,影响的Level分别获得LogLevel
- 获得logger对应的name,如果name等于null,或者等于root,则将其赋值为root
- 实例化LoggerConfiguration进行返回
- 如果 LoggerLevels,则返回ResponseEntity–>状态码为404,否则,直接返回LoggerLevels
set–>该方法用于设置logger的日志级别.代码如下:
@ActuatorPostMapping("/{name:.*}") @ResponseBody @HypermediaDisabled public Object set(@PathVariable String name, @RequestBody Map<String, String> configuration) { // 1. 如果不可用,则返回默认的不可用信息 if (!this.delegate.isEnabled()) { // Shouldn't happen - MVC endpoint shouldn't be registered when delegate's // disabled return getDisabledResponse(); } // 2. 根据configuration获得LogLevel,然后对指定的logger设置日志级别 LogLevel logLevel = getLogLevel(configuration); this.delegate.setLogLevel(name, logLevel); // 3. 返回ok return ResponseEntity.ok().build(); }
逻辑如下:
- 如果不可用,则返回默认的不可用信息
根据configuration获得LogLevel,代码如下:
private LogLevel getLogLevel(Map<String, String> configuration) { String level = configuration.get("configuredLevel"); try { return (level == null ? null : LogLevel.valueOf(level.toUpperCase())); } catch (IllegalArgumentException ex) { throw new InvalidLogLevelException(level); } }
- 获得configuredLevel对应的值
- 如果没有配置,则返回null,否则,返回LogLevel,合法值有TRACE, DEBUG, INFO, WARN, ERROR, FATAL, OFF.如果值不合法,抛出InvalidLogLevelException异常.由于InvalidLogLevelException类注有@ResponseStatus(value = HttpStatus.BAD_REQUEST, reason = “No such log level”),因此,当抛出该异常时,会返回相应的信息.
对指定的logger设置日志级别,代码如下:
public void setLogLevel(String name, LogLevel level) { Assert.notNull(name, "Name must not be empty"); this.loggingSystem.setLogLevel(name, level); }
默认情况下,会执行LogbackLoggingSystem#setLogLevel,代码如下:
public void setLogLevel(String loggerName, LogLevel level) { ch.qos.logback.classic.Logger logger = getLogger(loggerName); if (logger != null) { logger.setLevel(LEVELS.convertSystemToNative(level)); } }
- 根据对应的logger名获得Logger
- 将传入的loggerLevel转换为LogBack对应的日志级别
- 修改日志级别
- 返回ok
至此,我们明白,要想修改日志级别,可以通过对/loggers/logger名,发送post请求,传入类上如下格式的参数:
{"configuredLevel":"INFO"}
即可修改日志级别
参数配置–>因为该类有@ConfigurationProperties(prefix = “endpoints.loggers”)注解:
endpoints.loggers.path=/logfile # Endpoint path
自动化装配:
在EndpointWebMvcManagementContextConfiguration中进行了声明,代码如下:
@Bean @ConditionalOnMissingBean @ConditionalOnBean(LoggersEndpoint.class) @ConditionalOnEnabledEndpoint("loggers") public LoggersMvcEndpoint loggersMvcEndpoint(LoggersEndpoint delegate) { return new LoggersMvcEndpoint(delegate); }
- @Bean –> 注册1个id为loggersMvcEndpoint,类型为LoggersMvcEndpoint的bean
- @ConditionalOnMissingBean–> 当BeanFactory中不包含类型为LoggersMvcEndpoint的bean时生效
- @ConditionalOnBean(LoggersEndpoint.class)–> 当BeanFactory中存在LoggersEndpoint类型的bean时生效
- @ConditionalOnEnabledEndpoint(“loggers”)–> 如果配置有endpoints. loggers.enabled = true 或者endpoints.enabled= true 则该配置生效
ShutdownMvcEndpoint
该类继承自EndpointMvcAdapter
构造器如下:
public ShutdownMvcEndpoint(ShutdownEndpoint delegate) { super(delegate); }
invoke 实现如下:
@PostMapping(produces = { ActuatorMediaTypes.APPLICATION_ACTUATOR_V1_JSON_VALUE, MediaType.APPLICATION_JSON_VALUE }) @ResponseBody @Override public Object invoke() { if (!getDelegate().isEnabled()) { return new ResponseEntity<Map<String, String>>( Collections.singletonMap("message", "This endpoint is disabled"), HttpStatus.NOT_FOUND); } return super.invoke(); }
如果ShutdownEndpoint不可用,则返回{message:This endpoint is disabled},否则,调用ShutdownEndpoint#invoke,关闭spring boot 程序
参数配置–> 因为该类声明了@ConfigurationProperties(prefix = “endpoints.shutdown”):
endpoints.shutdown.path= # Endpoint path.
自动装配:
在EndpointWebMvcManagementContextConfiguration中进行了声明,代码如下:
@Bean @ConditionalOnMissingBean @ConditionalOnBean(ShutdownEndpoint.class) @ConditionalOnEnabledEndpoint(value = "shutdown", enabledByDefault = false) public ShutdownMvcEndpoint shutdownMvcEndpoint(ShutdownEndpoint delegate) { return new ShutdownMvcEndpoint(delegate); }
- @Bean –> 注册1个id为shutdownMvcEndpoint,类型为ShutdownMvcEndpoint的bean
- @ConditionalOnMissingBean–> 当BeanFactory中不包含类型为ShutdownMvcEndpoint的bean时生效
- @ConditionalOnBean(ShutdownEndpoint.class)–> 当BeanFactory中存在ShutdownEndpoint类型的bean时生效
- @ConditionalOnEnabledEndpoint(“shutdown”, enabledByDefault = false) –>如果配置有endpoints. shutdown.enabled = true则该配置生效,如果没有配置,该配置不生效
MetricsMvcEndpoint
MetricsMvcEndpoint–> 继承自EndpointMvcAdapter.
字段构造器如下:
private final MetricsEndpoint delegate; public MetricsMvcEndpoint(MetricsEndpoint delegate) { super(delegate); this.delegate = delegate; }
该类声明了1个被@ActuatorGetMapping注解的方法,如下:
@ActuatorGetMapping("/{name:.*}") @ResponseBody @HypermediaDisabled public Object value(@PathVariable String name) { if (!this.delegate.isEnabled()) { return getDisabledResponse(); } return new NamePatternMapFilter(this.delegate.invoke()).getResults(name); }
- 如果不可用,则返回默认的不可用信息
实例化NamePatternMapFilter ,之后调用其getResults方法,根据传入的name 获得对应的map,key –>name,value–>name所对应的Metric的值
NamePatternMapFilter:
- 继承了NamePatternFilter 接口,泛型参数为Map
抽象方法实现如下:
getValue–> 调用getOptionalValue获得MetricsEndpoint中Metrics名字为传入的值所对应的Metrics的值.,如果获取不到,则抛出NoSuchMetricException.代码如下:
protected Object getValue(Map<String, ?> source, String name) { Object value = getOptionalValue(source, name); if (value == null) { throw new NoSuchMetricException("No such metric: " + name); } return value; }
NoSuchMetricException代码如下:
@ResponseStatus(value = HttpStatus.NOT_FOUND, reason = "No such metric") public static class NoSuchMetricException extends RuntimeException { public NoSuchMetricException(String string) { super(string); } }
当抛出该异常时,返回的状态码为404,reason为No such metric
getOptionalValue–> 获得MetricsEndpoint中Metrics名字为传入的值所对应的Metrics的值.代码如下:
protected Object getOptionalValue(Map<String, ?> source, String name) { return source.get(name); }
getNames–>遍历MetricsEndpoint中返回的Metrics名,如果Metrics名符合正则的话(MetricsMvcEndpoint#value方法传入的是正则),则加入到ResultCollectingNameCallback的result中.代码如下:
protected void getNames(Map<String, ?> source, NameCallback callback) { for (String name : source.keySet()) { try { callback.addName(name); } catch (NoSuchMetricException ex) { // Metric with null value. Continue. } } }
自动装配:
声明在EndpointWebMvcManagementContextConfiguration中.代码如下:
@Bean @ConditionalOnMissingBean @ConditionalOnBean(MetricsEndpoint.class) @ConditionalOnEnabledEndpoint("metrics") public MetricsMvcEndpoint metricsMvcEndpoint(MetricsEndpoint delegate) { return new MetricsMvcEndpoint(delegate); }
- @Bean –> 注册1个id为metricsMvcEndpoint,类型为MetricsMvcEndpoint的bean
- @ConditionalOnMissingBean–> BeanFactory中不存在MetricsMvcEndpoint类型的bean时生效
- @ConditionalOnBean(MetricsEndpoint.class)–> BeanFactory中存在MetricsEndpoint类型的bean时生效
- @ConditionalOnEnabledEndpoint(“metrics”) –> 如果配置有endpoints. metrics.enabled = true 或者endpoints.enabled= true 则该配置生效