spring-boot-admin(以下简称SBA)与Spring Boot、Spring cloud项目以starter的方式自动集成，包括Server端和Client端。

SBA监控包括应用的基本信息、logfile(日志在线实时浏览或者download)、JVM信息(实时线程信息、堆信息、非堆信息)、Web(API接口信息、最近100次API调用的信息)、应用中用户登录信息;监控很全面、齐全，但部分监控指标与自己的项目需要还是有点差别的，比如以下两点：

自定义HttpTrace增加入参和出参

结果：

在spring-boot-admin中HttpTrace显示的信息包括session、principal、request、response、timeTaken和timestamp,但session、principal对我无用，request是HttpTrace的内部类显示的信息包括：

private final String method;private final URI uri;private final Map<String, List<String>> headers;private final String remoteAddress;

response也是HttpTrace的内部类显示的信息包括:

private final int status;private final Map<String, List<String>> headers;

唯一缺少的就是请求的请求入参和响应出参，而Headers的信息对该项目完全无用。所以势必要扩展HttpTrace才行,大致的思路是：自定义Filter-->装饰模式转换成自定义的request和response对象,内部获取请求和相应内容-->HttpExchangeTracer创建HttpTrace对象-->InmemoryHttpTraceRepository保存100次请求的HttpTrace对象，供server端使用。由于Filter中使用的部分对象需要先创建所以我们先从需要的零部件开始：

第一步:包装HttpServletRequest获取请求内容:

public class RequestWrapper extends HttpServletRequestWrapper { //存放请求的消息体(先缓存一份) private byte[] body; //自定义输入流的包装类，将缓存数据再写入到流中 private ServletInputStreamWrapper wrapper; private final Logger logger = LoggerFactory.getLogger(RequestWrapper.class); public RequestWrapper(HttpServletRequest request) { super(request); try { //使用Apache的commons-io工具从request中先读取数据 body = IOUtils.toByteArray(request.getInputStream()); } catch (IOException e) { logger.error("从请求中获取请求参数出现异常:", e); } //将读取出来的内存再写入流中 wrapper = new ServletInputStreamWrapper(new ByteArrayInputStream(body)); } //转换成String 供外部调用,并替换转义字符 public String body() { return new String(body).replaceAll("[\n\t\r]",""); } //将我们的自定义的流包装类返回，供系统调用 读取数据 @Override public ServletInputStream getInputStream() throws IOException { return this.wrapper; } //将我们的自定义的流包装类返回，供系统调用 读取数据 @Override public BufferedReader getReader() throws IOException { return new BufferedReader(new InputStreamReader(this.wrapper)); } //从给定的输入流中读取数据 static final class ServletInputStreamWrapper extends ServletInputStream { private InputStream inputStream; public ServletInputStreamWrapper(InputStream inputStream) { this.inputStream = inputStream; } @Override public boolean isFinished() { return true; } @Override public boolean isReady() { return false; } @Override public void setReadListener(ReadListener listener) { } @Override public int read() throws IOException { return this.inputStream.read(); } public InputStream getInputStream() { return inputStream; } public void setInputStream(InputStream inputStream) { this.inputStream = inputStream; } }}

第二步:包装HttpServletResponse类获取响应内容：

public class ResponseWrapper extends HttpServletResponseWrapper { private HttpServletResponse response; //缓存响应内容的输出流 private ByteArrayOutputStream result = new ByteArrayOutputStream(); public ResponseWrapper(HttpServletResponse response) { super(response); this.response = response; } /** * 响应的内容 供外部调用 *针对 体积较大的响应内容 很容易发生 OOM(比如：/actuator/logfile 接口)，可在调 用 *该方法的地方就行api过滤 *解决方法在第四步 */ public String body(){ return result.toString(); } @Override public ServletOutputStream getOutputStream() throws IOException { return new ServletOutputStreamWrapper(this.response,this.result); } @Override public PrintWriter getWriter() throws IOException {return new PrintWriter(new OutputStreamWriter(this.result,this.response.getCharacterEncoding())); }//自定义输出流 用于读出数据 static final class ServletOutputStreamWrapper extends ServletOutputStream{ private HttpServletResponse response; private ByteArrayOutputStream byteArrayOutputStream; public ServletOutputStreamWrapper(HttpServletResponse response, ByteArrayOutputStream byteArrayOutputStream) { this.response = response; this.byteArrayOutputStream = byteArrayOutputStream; } @Override public boolean isReady() { return true; } @Override public void setWriteListener(WriteListener listener) { } @Override public void write(int b) throws IOException { this.byteArrayOutputStream.write(b); } /** * 将内容重新刷新到返回的对象中 并且避免多次刷新 */ @Override public void flush() throws IOException { if(!response.isCommitted()){ byte[] bytes = this.byteArrayOutputStream.toByteArray(); ServletOutputStream outputStream = response.getOutputStream(); outputStream.write(bytes); outputStream.flush(); } } }}

第三步:扩展TraceableRequest,该接口中的方法会在创建HttpTrace#Request内部类时调用，自定义实现里面的方法，再在过滤器中引用该类就可以达到自定义监控内容的目的,该类中的Request是我们第一步创建的装饰类，不能使用HttpServletRequest

public class CustomerTraceableRequest implements TraceableRequest { //自定义的Request装饰类，不能使用HttpServletRequest private RequestWrapper request; public CustomerTraceableRequest(RequestWrapper request) { this.request = request; }//HttpTrace类中getMethod会调用 @Override public String getMethod() { return request.getMethod(); } /** * @return POST 或者 GET 方式 都返回 {ip}:{port}/uir的形式返回 */ @Override public URI getUri() { return URI.create(request.getRequestURL().toString()); }//因为在HttpTrace中可扩展的只有headers的Map,所以我们自定义属性RequestParam存入headers中，作为入参信息展示 @Override public Map<String, List<String>> getHeaders() { Map<String, List<String>> headerParam = new HashMap<>(1); headerParam.put("RequestParam",getParams()); return headerParam; }//该方法也要重写，默认的太简单无法获取真是的IP @Override public String getRemoteAddress() { return IpUtils.getIpAddress(request); }//根据GET或者POST的请求方式不同,获取不同情况下的请求参数 public List<String> getParams() { String params = null; String method = this.getMethod(); if(HttpMethod.GET.matches(method)){ params = request.getQueryString(); }else if(HttpMethod.POST.matches(method)){ params = this.request.body(); } List<String> result = new ArrayList<>(1); result.add(params); return result; }}

第四步:扩展TraceableResponse,该该接口中方法在创建HttpTrace#Response内部类时引用，自定义实现里面的方法,在端监控页面展示：

public class CustomerTraceableResponse implements TraceableResponse { //自定义HttpServletResponse包装类 不能使用HttpServletResponse对象 private ResponseWrapper response; private HttpServletRequest request; public CustomerTraceableResponse(ResponseWrapper response, HttpServletRequest request) { this.response = response; this.request = request; }//返回响应状态 @Override public int getStatus() { return response.getStatus(); }//扩展Response headers添加Response Body属性，展示响应内容，但是需要排除/actuator/开头的请求，这里面部分响应内容太大，容易OOM @Override public Map<String, List<String>> getHeaders() { if(isActuatorUri()){ return extractHeaders(); }else{ Map<String, List<String>> result = new LinkedHashMap<>(1); List<String> responseBody = new ArrayList<>(1); responseBody.add(this.response.body()); result.put("ResponseBody", responseBody); result.put("Content-Type", getContentType()); return result; } }//是否是需要过滤的请求uri private boolean isActuatorUri() { String requestUri = request.getRequestURI(); AntPathMatcher matcher = new AntPathMatcher(); return matcher.match("/actuator/**", requestUri); }//server端页面展示的Content-Type以及Length是从Response中获取的 private List<String> getContentType() { List<String> list = new ArrayList<>(1); list.add(this.response.getContentType()); return list; }//针对/actuator/**的请求返回默认的headers内容获 private Map<String, List<String>> extractHeaders() { Map<String, List<String>> headers = new LinkedHashMap<>(); for (String name : this.response.getHeaderNames()) { headers.put(name, new ArrayList<>(this.response.getHeaders(name))); } return headers; }}

第五步:自定义Filter对Resquest和Response过滤,并创建HttpTrace对象并保存：

public class CustomerHttpTraceFilter extends OncePerRequestFilter implements Ordered {//该类存储HttpTrace的repository,默认是基于内存的，可扩展该类跟换存储数据的方式 private HttpTraceRepository httpTraceRepository;//该类创建HttpTrace对象，通过Set<Include>在配置文件中我们需要展示那些内容的容器(request-headers,response-headers,remote-address,time-taken) 对request和response进行筛选 private HttpExchangeTracer httpExchangeTracer; public CustomerHttpTraceFilter(HttpTraceRepository httpTraceRepository, HttpExchangeTracer httpExchangeTracer) { this.httpTraceRepository = httpTraceRepository; this.httpExchangeTracer = httpExchangeTracer; } @Override protected void doFilterInternal(HttpServletRequest request, HttpServletResponse response, FilterChain filterChain) throws ServletException, IOException { //校验URI是否有效 if (!isRequestValid(request)) { filterChain.doFilter(request, response); return; } //将HttpServletRequest包装成我们自己的 RequestWrapper wrapper = new RequestWrapper(request); //将HttpServletResponse包装成我们的自己的 ResponseWrapper responseWrapper = new ResponseWrapper(response); //创建我们的自己的TraceRequest对象 CustomerTraceableRequest traceableRequest = new CustomerTraceableRequest(wrapper);//创建HttpTrace对象(FilteredTraceableRequest 是内部类，通过Set<Include>筛选那些信息需要展示就保存那些信息),重点设置HttpTrace#Request对象的各种参数 HttpTrace httpTrace = httpExchangeTracer.receivedRequest(traceableRequest); try { filterChain.doFilter(wrapper, responseWrapper); } finally {//使用自定义的TraceableResponse 保存需要的response信息 CustomerTraceableResponse traceableResponse = new CustomerTraceableResponse(responseWrapper,request); //根据Set<Include>设置HttpTrace中session、principal、timeTaken信息以及 Response内部类信息  this.httpExchangeTracer.sendingResponse(httpTrace, traceableResponse, null, null); //将HttpTrace对象保存在Respository中存储起来 this.httpTraceRepository.add(httpTrace); } } private boolean isRequestValid(HttpServletRequest request) { try { new URI(request.getRequestURL().toString()); return true; } catch (URISyntaxException ex) { return false; } } @Override public int getOrder() { return Ordered.LOWEST_PRECEDENCE - 10; }}

第六步:通过@SpringBootApplication(exclude)禁用HttpTraceAutoConfiguration自动配置，自定义自动配置更换Filter过滤器：

@Configuration@ConditionalOnWebApplication@ConditionalOnProperty(prefix = "management.trace.http", name = "enabled", matchIfMissing = true)@EnableConfigurationProperties(HttpTraceProperties.class)public class TraceFilterConfig { //存储HttpTrace信息的对象 @Bean @ConditionalOnMissingBean(HttpTraceRepository.class) public InMemoryHttpTraceRepository traceRepository() { return new InMemoryHttpTraceRepository(); } //创建HttpTrace对象Exchange @Bean @ConditionalOnMissingBean public HttpExchangeTracer httpExchangeTracer(HttpTraceProperties traceProperties) { return new HttpExchangeTracer(traceProperties.getInclude()); } @ConditionalOnWebApplication(type = ConditionalOnWebApplication.Type.SERVLET) static class ServletTraceFilterConfiguration { //将我们自定义的Filter已Bean的方式注册，才能生效 @Bean @ConditionalOnMissingBean public CustomerHttpTraceFilter httpTraceFilter(HttpTraceRepository repository, HttpExchangeTracer tracer) { return new CustomerHttpTraceFilter(repository,tracer); } } @ConditionalOnWebApplication(type = ConditionalOnWebApplication.Type.REACTIVE) static class ReactiveTraceFilterConfiguration { @Bean @ConditionalOnMissingBean public HttpTraceWebFilter httpTraceWebFilter(HttpTraceRepository repository, HttpExchangeTracer tracer, HttpTraceProperties traceProperties) { return new HttpTraceWebFilter(repository, tracer, traceProperties.getInclude()); } } }

集成Redisson健康状态监控

如果有引入spring-boot-starter-redis,SBA默认同过RedisConnectionFactory监控Redis的健康状态，无奈Redisson还没有，自己动手丰衣足食。通过HealthIndicator或ReactiveHealthIndicator使用策略模式实现不同组件的健康监控，后者是使用Rective模式下的。我是通过JavaBean的方式配置Redisson,顺便再添加ReactiveHealthIndicator该指标即可：

@Configuration@EnableConfigurationProperties(value = RedissonProperties.class)public class RedissonConfig implements ReactiveHealthIndicator { //自己的RedissonProperties文件 @Autowired private RedissonProperties redissonProperties; //暴露 redissonClient句柄 @Bean @ConditionalOnMissingBean public RedissonClient redisClient() { return Redisson.create(config()); } //通过Bean的方式配置RedissonConfig相关信息 @Bean public Config config() { Config config = new Config(); config.useSingleServer() //单实列模式 .setAddress(redissonProperties.getAddress() + ":" + redissonProperties.getPort()) .setPassword(redissonProperties.getPassword()) .setDatabase(redissonProperties.getDatabase()) .setConnectionPoolSize(redissonProperties.getConnectionPoolSize()) .setConnectionMinimumIdleSize(redissonProperties.getConnectionMinimumIdleSize()) .setIdleConnectionTimeout(redissonProperties.getIdleConnectionTimeout()) .setSubscriptionConnectionPoolSize(redissonProperties.getSubscriptionConnectionPoolSize()) .setSubscriptionConnectionMinimumIdleSize(redissonProperties.getSubscriptionConnectionMinimumIdleSize()) .setTimeout(redissonProperties.getTimeout()) .setRetryAttempts(redissonProperties.getRetryAttempts()) .setRetryInterval(redissonProperties.getRetryInterval()) .setConnectTimeout(redissonProperties.getConnectTimeout()) .setReconnectionTimeout(redissonProperties.getReconnectionTimeout()); config .setCodecProvider(new DefaultCodecProvider()) .setEventLoopGroup(new NioEventLoopGroup()) .setThreads(Runtime.getRuntime().availableProcessors() * 2) .setNettyThreads(Runtime.getRuntime().availableProcessors() * 2); return config; } //实现ReactiveHealthIndicator 重写health方法 @Override public Mono<Health> health() { return checkRedissonHealth().onErrorResume(ex -> Mono.just(new Health.Builder().down(ex).build())); } //我是通过ping 的方式判断redis服务器是否up的状态，并增加加Netty和Threads的监控 private Mono<Health> checkRedissonHealth() { Health.Builder builder = new Health.Builder(); builder.withDetail("address", redissonProperties.getAddress()); //检测健康状态 if (this.redisClient().getNodesGroup().pingAll()) { builder.status(Status.UP); builder.withDetail("dataBase", redissonProperties.getDatabase()); builder.withDetail("redisNodeThreads", this.redisClient().getConfig().getThreads()); builder.withDetail("nettyThreads", this.redisClient().getConfig().getNettyThreads()); }else{ builder.status(Status.DOWN); } return Mono.just(builder.build()); }}

在页面上看就是:

Ok！圆满完成！

如有错误，不吝赐教！

欢迎关注技术文章公众号：