Jdk8 DNS解析
注:JDK7和JDK8关于DNS解析的实现有差异,该问题在JDK7下可能不存在;
Java中的DNS解析一般是通过调用下面的方法:
public static InetAddress getByName(String host)
public static InetAddress[] getAllByName(String host)
getByName先调用getAllByName,然后返回地址列表的第一个地址;
下面主要看看getAllByName的实现;
getAllByName
getAllByName会调用getAllByName0方法:
InetAddress[] addresses = getCachedAddresses(host); /* If no entry in cache, then do the host lookup */
if (addresses == null) {
addresses = getAddressesFromNameService(host, reqAddr);
} if (addresses == unknown_array)
throw new UnknownHostException(host); return addresses.clone();
可以看到首先会从缓存中获取,如果缓存找不到则调用getAddressesFromNameService进行解析;
private static InetAddress[] getCachedAddresses(String hostname) {
hostname = hostname.toLowerCase(); // search both positive & negative caches synchronized (addressCache) {
cacheInitIfNeeded();//如果是第一次调用,执行初始化 CacheEntry entry = addressCache.get(hostname);
if (entry == null) {
entry = negativeCache.get(hostname);
} if (entry != null) {
return entry.addresses;
}
} // not found
return null;
}
既然JDK对IP地址解析有缓存,那么它是如何缓存的呢?缓存策略定义在InetAddressCachePolicy类,摘录其初始化的代码如下:
static {
Integer tmp = java.security.AccessController.doPrivileged(
new PrivilegedAction<Integer>() {
public Integer run() {
try {
//读取JDK目录java.security文件的属性networkaddress.cache.ttl
String tmpString = Security.getProperty(cachePolicyProp);
if (tmpString != null) {
return Integer.valueOf(tmpString);
}
} catch (NumberFormatException ignored) {
// Ignore
}
try {
//读取-D指定的系统属性sun.net.inetaddr.ttl
String tmpString = System.getProperty(cachePolicyPropFallback);
if (tmpString != null) {
return Integer.decode(tmpString);
}
} catch (NumberFormatException ignored) {
// Ignore
}
return null;
}
}); if (tmp != null) {
cachePolicy = tmp.intValue();
if (cachePolicy < 0) {
cachePolicy = FOREVER;//如果配置的是负数,表示缓存永不过期
}
propertySet = true;
} else {
//可以通过-Djava.security.manager-Djava.security.policy=security.policy启动安全管理器
if (System.getSecurityManager() == null) {
cachePolicy = DEFAULT_POSITIVE;//默认是不启动SecurityManager的,也就是说默认缓存失效时间为30s
}
}
tmp = java.security.AccessController.doPrivileged (
new PrivilegedAction<Integer>() {
public Integer run() {
try {
//读取networkaddress.cache.negative.ttl属性,默认是10s
String tmpString = Security.getProperty(negativeCachePolicyProp);
if (tmpString != null) {
return Integer.valueOf(tmpString);
}
} catch (NumberFormatException ignored) {
// Ignore
} try {
//读取-D指定的系统属性sun.net.inetaddr.negative.ttl
String tmpString = System.getProperty(negativeCachePolicyPropFallback);
if (tmpString != null) {
return Integer.decode(tmpString);
}
} catch (NumberFormatException ignored) {
// Ignore
}
return null;
}
}); if (tmp != null) {
negativeCachePolicy = tmp.intValue();
if (negativeCachePolicy < 0) {
negativeCachePolicy = FOREVER;
}
propertyNegativeSet = true;
}
}
上面介绍了JVM对ip地址解析的缓存策略和相关的配置,接下来看看,如果缓存找不到,JVM该如何解析ip地址;
getAddressesFromNameService
从上面的代码看到,InetAddress会调用getAddressesFromNameService方法,循环调用nameService的lookupAllHostAddr方法,直到找到结果:
NameService的初始化代码如下:
impl = InetAddressImplFactory.create(); // get name service if provided and requested
String provider = null;;
String propPrefix = "sun.net.spi.nameservice.provider.";
int n = 1;
nameServices = new ArrayList<NameService>();
//可以通过sun.net.spi.nameservice.provider.n指定自己的DNS
Provider
provider = AccessController.doPrivileged(
new GetPropertyAction(propPrefix + n));
while (provider != null) {
NameService ns = createNSProvider(provider);
if (ns != null)
nameServices.add(ns); n++;
provider = AccessController.doPrivileged(
new GetPropertyAction(propPrefix + n));
} //如果不单独指定,创建默认的NameService
if (nameServices.size() == 0) {//
NameService ns = createNSProvider("default");
nameServices.add(ns);
}
在这里要特别提下Java提供的DNSNameService,该类可以通过下述参数启用:
-Dsun.net.spi.nameservice.provider.1=dns,sun
-Dsun.net.spi.nameservice.nameservers=192.168.1.188
该类会根据sun.net.spi.nameservice.nameservers指定的name server或/etc/resolv.conf文件中配置的name server进行DNS解析;
创建默认的NameService方法代码如下:
if (provider.equals("default")) {
// initialize the default name service
nameService = new NameService() {
public InetAddress[] lookupAllHostAddr(String host)
throws UnknownHostException {
return impl.lookupAllHostAddr(host);
}
public String getHostByAddr(byte[] addr)
throws UnknownHostException {
return impl.getHostByAddr(addr);
}
};
}
根据指定的provider创建NameService的方法如下:
nameService = java.security.AccessController.doPrivileged(
new java.security.PrivilegedExceptionAction<NameService>() {
public NameService run() {
Iterator itr = Service.providers(NameServiceDescriptor.class);
while (itr.hasNext()) {
NameServiceDescriptor nsd
= (NameServiceDescriptor)itr.next();
if (providerName.
equalsIgnoreCase(nsd.getType()+","
+nsd.getProviderName())) {
try {
return nsd.createNameService();
} catch (Exception e) {
e.printStackTrace();
System.err.println(
"Cannot create name service:"
+providerName+": " + e);
}
}
} return null;
}
}
);
对于DNSNameServiceDescriptor,其Type和ProviderName分别为dns,sun;
继续看默认Provider的处理逻辑,可以看到其是通过impl.lookupAllHostAddr(host)方法进行解析的,impl的初始化代码为:
impl = InetAddressImplFactory.create();
static InetAddressImpl create() {
return InetAddress.loadImpl(isIPv6Supported() ?
"Inet6AddressImpl" : "Inet4AddressImpl");
}
这里以Inet4AddressImpl为例,说明DNS的解析:
public native InetAddress[]
lookupAllHostAddr(String hostname) throws UnknownHostException;
public native String getHostByAddr(byte[] addr) throws UnknownHostException;
Inet4AddressImp类的方法是native的,是采用本地方法实现的:
JNIEXPORT jobjectArray JNICALL
Java_java_net_Inet4AddressImpl_lookupAllHostAddr(JNIEnv *env, jobject this,
jstring host) {
const char *hostname;
jobjectArray ret = 0;
int retLen = 0;
int error = 0;
struct addrinfo hints, *res, *resNew = NULL; if (!initializeInetClasses(env))
return NULL; if (IS_NULL(host)) {
JNU_ThrowNullPointerException(env, "host is null");
return 0;
}
hostname = JNU_GetStringPlatformChars(env, host, JNI_FALSE);
CHECK_NULL_RETURN(hostname, NULL); /* Try once, with our static buffer. */
memset(&hints, 0, sizeof(hints));
hints.ai_flags = AI_CANONNAME;
hints.ai_family = AF_INET; error = getaddrinfo(hostname, NULL, &hints, &res); if (error) {
/* report error */
ThrowUnknownHostExceptionWithGaiError(env, hostname, error);
JNU_ReleaseStringPlatformChars(env, host, hostname);
return NULL;
} else {
int i = 0;
struct addrinfo *itr, *last = NULL, *iterator = res; while (iterator != NULL) {
// remove the duplicate one
int skip = 0;
itr = resNew;
while (itr != NULL) {
struct sockaddr_in *addr1, *addr2;
addr1 = (struct sockaddr_in *)iterator->ai_addr;
addr2 = (struct sockaddr_in *)itr->ai_addr;
if (addr1->sin_addr.s_addr ==
addr2->sin_addr.s_addr) {
skip = 1;
break;
}
itr = itr->ai_next;
} if (!skip) {
struct addrinfo *next
= (struct addrinfo*) malloc(sizeof(struct addrinfo));
if (!next) {
JNU_ThrowOutOfMemoryError(env, "Native heap allocation failed");
ret = NULL;
goto cleanupAndReturn;
}
memcpy(next, iterator, sizeof(struct addrinfo));
next->ai_next = NULL;
if (resNew == NULL) {
resNew = next;
} else {
last->ai_next = next;
}
last = next;
i++;
}
iterator = iterator->ai_next;
} retLen = i;
iterator = resNew; ret = (*env)->NewObjectArray(env, retLen, ni_iacls, NULL); if (IS_NULL(ret)) {
/* we may have memory to free at the end of this */
goto cleanupAndReturn;
} i = 0;
while (iterator != NULL) {
jobject iaObj = (*env)->NewObject(env, ni_ia4cls, ni_ia4ctrID);
if (IS_NULL(iaObj)) {
ret = NULL;
goto cleanupAndReturn;
}
setInetAddress_addr(env, iaObj, ntohl(((struct sockaddr_in*)iterator->ai_addr)->sin_addr.s_addr));
setInetAddress_hostName(env, iaObj, host);
(*env)->SetObjectArrayElement(env, ret, i++, iaObj);
iterator = iterator->ai_next;
}
} }
上面的代码一大堆,核心是调用getaddrinfo函数,在getaddrinfo的man文档中有这么一句话:
the application should try using the addresses in the order in which they are returned. The sorting function used within getaddrinfo() is defined in RFC 3484; the order can be tweaked for a
particular system by editing /etc/gai.conf (available since glibc 2.5).
getaddrinfo返回的地址列表根据RFC3484规定的排序算法进行了排序,如果这样的话,那么返回的地址列表顺序是规定的,那就达不到负载均衡的目的了;
关于这个排序的话题,网上有很多讨论:
- https://lists.debian.org/debian-glibc/2007/09/msg00347.html
- https://lists.debian.org/debian-ctte/2007/09/msg00067.html
- https://daniel.haxx.se/blog/2012/01/03/getaddrinfo-with-round-robin-dns-and-happy-eyeballs/
getaddrinfo的部分代码如下:
int getaddrinfo (const char *__restrict name, const char *__restrict service,
const struct addrinfo *__restrict hints,
struct addrinfo **__restrict pai)
{
int i = 0, j = 0, last_i = 0;
int nresults = 0;
struct addrinfo *p = NULL, **end;
struct gaih *g = gaih, *pg = NULL;
struct gaih_service gaih_service, *pservice;
struct addrinfo local_hints; while (g->gaih)
{
if (hints->ai_family == g->family || hints->ai_family == AF_UNSPEC)
{
j++;
if (pg == NULL || pg->gaih != g->gaih)
{
pg = g;
i = g->gaih (name, pservice, hints, end);
if (i != 0)
{
/* EAI_NODATA is a more specific result as it says that
we found a result but it is not usable. */
if (last_i != (GAIH_OKIFUNSPEC | -EAI_NODATA))
last_i = i;
if (hints->ai_family == AF_UNSPEC && (i & GAIH_OKIFUNSPEC))
{
++g;
continue;
}
freeaddrinfo (p);
return -(i & GAIH_EAI);
}
if (end)
while (*end)
{
end = &((*end)->ai_next);
++nresults;
}
}
}
++g;
}
if (j == 0)
return EAI_FAMILY;
if (nresults > 1)
{
/* Sort results according to RFC 3484. */
struct sort_result results[nresults];
struct addrinfo *q;
struct addrinfo *last = NULL;
char *canonname = NULL;
for (i = 0, q = p; q != NULL; ++i, last = q, q = q->ai_next)
{
results[i].dest_addr = q;
results[i].got_source_addr = false;
/* If we just looked up the address for a different
protocol, reuse the result. */
if (last != NULL && last->ai_addrlen == q->ai_addrlen
&& memcmp (last->ai_addr, q->ai_addr, q->ai_addrlen) == 0)
{
memcpy (&results[i].source_addr, &results[i - 1].source_addr,
results[i - 1].source_addr_len);
results[i].source_addr_len = results[i - 1].source_addr_len;
results[i].got_source_addr = results[i - 1].got_source_addr;
}
else
{
/* We overwrite the type with SOCK_DGRAM since we do not
want connect() to connect to the other side. If we
cannot determine the source address remember this
fact. */
int fd = socket (q->ai_family, SOCK_DGRAM, IPPROTO_IP);
socklen_t sl = sizeof (results[i].source_addr);
if (fd != -1
&& connect (fd, q->ai_addr, q->ai_addrlen) == 0
&& getsockname (fd,
(struct sockaddr *) &results[i].source_addr,
&sl) == 0)
{
results[i].source_addr_len = sl;
results[i].got_source_addr = true;
}
else
/* Just make sure that if we have to process the same
address again we do not copy any memory. */
results[i].source_addr_len = 0;
if (fd != -1)
close_not_cancel_no_status (fd);
}
/* Remember the canonical name. */
if (q->ai_canonname != NULL)
{
assert (canonname == NULL);
canonname = q->ai_canonname;
q->ai_canonname = NULL;
}
}
/* We got all the source addresses we can get, now sort using
the information. */
qsort (results, nresults, sizeof (results[0]), rfc3484_sort);
/* Queue the results up as they come out of sorting. */
q = p = results[0].dest_addr;
for (i = 1; i < nresults; ++i)
q = q->ai_next = results[i].dest_addr;
q->ai_next = NULL;
/* Fill in the canonical name into the new first entry. */
p->ai_canonname = canonname;
}
if (p)
{
*pai = p;
return 0;
}
if (pai == NULL && last_i == 0)
return 0;
return last_i ? -(last_i & GAIH_EAI) : EAI_NONAME;
}
排序是通过rfc3484_sort完成的,后面有时间准备仔细看看其排序规则:
static int
rfc3484_sort (const void *p1, const void *p2)
{
const struct sort_result *a1 = (const struct sort_result *) p1;
const struct sort_result *a2 = (const struct sort_result *) p2;
/* Rule 1: Avoid unusable destinations.
We have the got_source_addr flag set if the destination is reachable. */
if (a1->got_source_addr && ! a2->got_source_addr)
return -1;
if (! a1->got_source_addr && a2->got_source_addr)
return 1;
/* Rule 2: Prefer matching scope. Only interesting if both
destination addresses are IPv6. */
int a1_dst_scope
= get_scope ((struct sockaddr_storage *) a1->dest_addr->ai_addr);
int a2_dst_scope
= get_scope ((struct sockaddr_storage *) a2->dest_addr->ai_addr);
if (a1->got_source_addr)
{
int a1_src_scope = get_scope (&a1->source_addr);
int a2_src_scope = get_scope (&a2->source_addr);
if (a1_dst_scope == a1_src_scope && a2_dst_scope != a2_src_scope)
return -1;
if (a1_dst_scope != a1_src_scope && a2_dst_scope == a2_src_scope)
return 1;
}
/* Rule 3: Avoid deprecated addresses.
That's something only the kernel could decide. */
/* Rule 4: Prefer home addresses.
Another thing only the kernel can decide. */
/* Rule 5: Prefer matching label. */
if (a1->got_source_addr)
{
int a1_dst_label
= get_label ((struct sockaddr_storage *) a1->dest_addr->ai_addr);
int a1_src_label = get_label (&a1->source_addr);
int a2_dst_label
= get_label ((struct sockaddr_storage *) a2->dest_addr->ai_addr);
int a2_src_label = get_label (&a2->source_addr);
if (a1_dst_label == a1_src_label && a2_dst_label != a2_src_label)
return -1;
if (a1_dst_label != a1_src_label && a2_dst_label == a2_src_label)
return 1;
}
/* Rule 6: Prefer higher precedence. */
int a1_prec
= get_precedence ((struct sockaddr_storage *) a1->dest_addr->ai_addr);
int a2_prec
= get_precedence ((struct sockaddr_storage *) a2->dest_addr->ai_addr);
if (a1_prec > a2_prec)
return -1;
if (a1_prec < a2_prec)
return 1;
/* Rule 7: Prefer native transport.
XXX How to recognize tunnels? */
/* Rule 8: Prefer smaller scope. */
if (a1_dst_scope < a2_dst_scope)
return -1;
if (a1_dst_scope > a2_dst_scope)
return 1;
/* Rule 9: Use longest matching prefix. */
if (a1->got_source_addr
&& a1->dest_addr->ai_family == a2->dest_addr->ai_family)
{
int bit1 = 0;
int bit2 = 0;
if (a1->dest_addr->ai_family == PF_INET)
{
assert (a1->source_addr.ss_family == PF_INET);
assert (a2->source_addr.ss_family == PF_INET);
struct sockaddr_in *in1_dst;
struct sockaddr_in *in1_src;
struct sockaddr_in *in2_dst;
struct sockaddr_in *in2_src;
in1_dst = (struct sockaddr_in *) a1->dest_addr->ai_addr;
in1_src = (struct sockaddr_in *) &a1->source_addr;
in2_dst = (struct sockaddr_in *) a2->dest_addr->ai_addr;
in2_src = (struct sockaddr_in *) &a2->source_addr;
bit1 = ffs (in1_dst->sin_addr.s_addr ^ in1_src->sin_addr.s_addr);
bit2 = ffs (in2_dst->sin_addr.s_addr ^ in2_src->sin_addr.s_addr);
}
else if (a1->dest_addr->ai_family == PF_INET6)
{
assert (a1->source_addr.ss_family == PF_INET6);
assert (a2->source_addr.ss_family == PF_INET6);
struct sockaddr_in6 *in1_dst;
struct sockaddr_in6 *in1_src;
struct sockaddr_in6 *in2_dst;
struct sockaddr_in6 *in2_src;
in1_dst = (struct sockaddr_in6 *) a1->dest_addr->ai_addr;
in1_src = (struct sockaddr_in6 *) &a1->source_addr;
in2_dst = (struct sockaddr_in6 *) a2->dest_addr->ai_addr;
in2_src = (struct sockaddr_in6 *) &a2->source_addr;
int i;
for (i = 0; i < 4; ++i)
if (in1_dst->sin6_addr.s6_addr32[i]
!= in1_src->sin6_addr.s6_addr32[i]
|| (in2_dst->sin6_addr.s6_addr32[i]
!= in2_src->sin6_addr.s6_addr32[i]))
break;
if (i < 4)
{
bit1 = ffs (in1_dst->sin6_addr.s6_addr32[i]
^ in1_src->sin6_addr.s6_addr32[i]);
bit2 = ffs (in2_dst->sin6_addr.s6_addr32[i]
^ in2_src->sin6_addr.s6_addr32[i]);
}
}
if (bit1 > bit2)
return -1;
if (bit1 < bit2)
return 1;
}
/* Rule 10: Otherwise, leave the order unchanged. */
return 0;
}
可以看到,首先根据RFC3484的Rule1~Rule9排序,如果上述规则都未触发,则返回原列表;简单的说,返回结果的顺序是不固定的,有可能是DNS Server返回的顺序,也有可能不是;因此最好的办法是在Java层自己进行控制;
转自:https://www.jianshu.com/p/f10808ae4b60
作者:allanYan
Jdk8 DNS解析的更多相关文章
- DNS解析过程详解
先说一下DNS的几个基本概念: 一. 根域 就是所谓的“.”,其实我们的网址www.baidu.com在配置当中应该是www.baidu.com.(最后有一点),一般我们在浏览器里输入时会省略后面的点 ...
- C#实现DNS解析服务和智能DNS服务
C#实现DNS解析服务有一个开源项目ARSoft.Tools.Net, ARSoft.Tools.Net是一个非常强大的开源DNS控件库,包含.Net SPF validation, SenderID ...
- 解决域名DNS解析的故障
在实际应用过程中可能会遇到DNS解析错误的问题,就是说当我们访问一个域名时无法完成将其解析到IP地址的工作,而直接输入网站IP却可以正常访问,这就是因为DNS解析出现故障造成的.这个现象发生的机率比较 ...
- LINUX DNS解析的3种修改方法~
1.HOST 本地DNS解析 vi /etc/hosts 添加规则 例如: 223.231.234.33 www.baidu.com 2.网卡配置文件DNS服务地址 vi /etc/sysconfi ...
- DNS解析过程和域名收敛、域名发散、SPDY应用
前段时间项目要做域名收敛,糊里糊涂的完成了,好多原理不清晰,现在整理搜集下知识点. 域名收敛的目的是什么?简单来说就是域名解析慢.那为什么解析慢?且听下文慢慢道来. 什么是DNS? DNS( Doma ...
- 通过统计用户DNS解析记录,实现监控用户上网行为
上次通过扫描抓包分析TTL的方式检测公司网络开放的端口,发现没有开放53端口(DNS),也就是在公司内部的主机只能用服务器自动分配的DNS,并且发现这是台内部服务器.今天发现bing上不去,检测后发现 ...
- DNS解析全过程及原理
DNS解析原理及过程. 当用户访问我们网站一个网页时,他需要经过以下步骤: 1)找到这个网页的存放服务器: 2)服务器将用户的请求信息接入: 3)服务器通过文件路径(URL)查找用户请求网页: 4)用 ...
- 使用dig查询dns解析
原文地址:使用dig查询dns解析 作者:chenwenming 一般来说linux下查询域名解析有两种选择,nslookup或者dig,而在使用上我觉得dig更加方便顺手. 如果是在debian下的 ...
- Dnsmasq安装与配置-搭建本地DNS服务器 更干净更快无广告DNS解析
默认的情况下,我们平时上网用的本地DNS服务器都是使用电信或者联通的,但是这样也导致了不少的问题,首当其冲的就是上网时经常莫名地弹出广告,或者莫名的流量被消耗掉导致网速变慢.其次是部分网站域名不能正常 ...
随机推荐
- 小tips:TCP的三次握手、长连接、 短连接、 SPDY 协议
当网络通信时采用TCP协议时,在真正的读写操作之前,server与client之间必须建立一个连接,当读写操作完成后,双方不再需要这个连接时它们可以释放这个连接,连接的建立是需要三次握手的,而释放则需 ...
- Springboot jpa多数据源
1.SpringBootApplication package com.xx.xxx; import org.springframework.beans.factory.annotation.Auto ...
- Vue组件开发分享
在开始本文之前,你可能需要先了解以下相关内容: Vue.js 一款高性能轻量化的MVVM框架 Webpack 前端模块化代码构建工具 Vue组件介绍 基于vue.js高效的双向数据绑定特性,让我们 ...
- can解析
- Linux装B命令
原文:https://mp.weixin.qq.com/s/CNmMRjl0iZ8EBPq5VgJHsA 1.空心字体 yum install -y figlet figlet happy 1.0 2 ...
- day 33
目录 数据库是什么 为什么使用数据库 数据库的分类 关系型(把数据保存在硬盘里) 非关系型(把数据保存在内存里) mysql的架构 初识mysql 操作数据库 增 删 改 查 数据库是什么 数据库即存 ...
- kill 命令在Java应用中使用注意事项
前言 我们都知道,kill在linux系统中是用于杀死进程. kill pid [..] kill命令可将指定的信号发送给相应的进程或工作. kill命令默认使用信号为15,用于结束进程或工作.如果进 ...
- zabbix--显示插件Graphtree
zabbix配置显示插件Graphtree 网盘下载地址:https://pan.baidu.com/s/1HGALOq2z7XiVRxyb6ASAeQ 提取码:13vn Graphtree 安装 ...
- (1)jmeter录制脚本(使用代理的方式)
(1)jmeter录制脚本(使用代理的方式) jmeter 2018年07月09日 17时27分24秒 很多APP使用badboy是无法录制的,这种情况下需要使用chrome或Firefox,如果能联 ...
- SpringCloud2.0 Hystrix Feign 基于Feign实现断路器 基础教程(七)
1.启动[服务中心]集群,工程名:springcloud-eureka-server 参考 SpringCloud2.0 Eureka Server 服务中心 基础教程(二) 2.启动[服务提供者]集 ...