lua开发--web页面服务

来源:互联网 发布:车辆数据 编辑:程序博客网 时间:2024/06/05 23:44

这里写图片描述

技术选型

MQ可以使用如Apache ActiveMQ;

Worker/动态服务可以通过如Java技术实现;

RPC可以选择如alibaba Dubbo;

KV持久化存储可以选择SSDB(如果使用SSD盘则可以选择SSDB+RocksDB引擎)或者ARDB(LMDB引擎版);

缓存使用Redis;

SSDB/Redis分片使用如Twemproxy,这样不管使用Java还是Nginx+Lua,它们都不关心分片逻辑;

前端模板拼装使用Nginx+Lua;

数据集群数据存储的机器可以采用RAID技术或者主从模式防止单点故障;

因为数据变更不频繁,可以考虑SSD替代机械硬盘。

核心流程

1、首先我们监听商品数据变更消息;

2、接收到消息后,数据聚合Worker通过RPC调用相关系统获取所有要展示的数据,此处获取数据的来源可能非常多而且响应速度完全受制于这些系统,可能耗时几百毫秒甚至上秒的时间;

3、将数据聚合为JSON串存储到相关数据集群;

4、前端Nginx通过Lua获取相关集群的数据进行展示;商品页需要获取基本信息+其他信息进行模板拼装,即拼装模板仅需要两次调用(另外因为其他信息数据量少且对一致性要求不高,因此我们完全可以缓存到Nginx本地全局内存,这样可以减少远程调用提高性能);当页面滚动到商品介绍页面时异步调用商品介绍服务获取数据;

5、如果从聚合的SSDB集群/Redis中获取不到相关数据;则回源到动态服务通过RPC调用相关系统获取所有要展示的数据返回(此处可以做限流处理,因为如果大量请求过来的话可能导致服务雪崩,需要采取保护措施),此处的逻辑和数据聚合Worker完全一样;然后发送MQ通知数据变更,这样下次访问时就可以从聚合的SSDB集群/Redis中获取数据了。

基本流程如上所述,主要分为Worker、动态服务、数据存储和前端展示;因为系统非常复杂,只介绍动态服务和前端展示、数据存储架构;Worker部分不做实现。

项目搭建

项目部署目录结构

/usr/chapter7  ssdb_basic_7770.conf  ssdb_basic_7771.conf  ssdb_basic_7772.conf  ssdb_basic_7773.conf  ssdb_desc_8880.conf  ssdb_desc_8881.conf  ssdb_desc_8882.conf  ssdb_desc_8883.conf  redis_other_6660.conf  redis_other_6661.conf  nginx_chapter7.conf  nutcracker.yml  nutcracker.init  item.html  header.html  footer.html  item.lua  desc.lua  lualib    item.lua    item      common.lua  webappWEB-INF   lib   classes   web.xml

数据存储实现

整体架构为主从模式,写数据到主集群,读数据从从集群读取数据,这样当一个集群不足以支撑流量时可以使用更多的集群来支撑更多的访问量;集群分片使用Twemproxy实现。

商品基本信息SSDB集群配置

vim /usr/chapter7/ssdb_basic_7770.confwork_dir = /usr/data/ssdb_7770pidfile = /usr/data/ssdb_7770.pidserver:        ip: 0.0.0.0        port: 7770        allow: 127.0.0.1        allow: 192.168replication:        binlog: yes        sync_speed: -1        slaveof:logger:        level: error        output: /usr/data/ssdb_7770.log        rotate:                size: 1000000000leveldb:        cache_size: 500        block_size: 32        write_buffer_size: 64        compaction_speed: 1000        compression: yes

vim /usr/chapter7/ssdb_basic_7771.conf

work_dir = /usr/data/ssdb_7771pidfile = /usr/data/ssdb_7771.pidserver:        ip: 0.0.0.0        port: 7771        allow: 127.0.0.1        allow: 192.168replication:        binlog: yes        sync_speed: -1        slaveof:logger:        level: error        output: /usr/data/ssdb_7771.log        rotate:                size: 1000000000leveldb:        cache_size: 500        block_size: 32        write_buffer_size: 64        compaction_speed: 1000        compression: yes

vim /usr/chapter7/ssdb_basic_7773.conf

work_dir = /usr/data/ssdb_7773pidfile = /usr/data/ssdb_7773.pidserver:        ip: 0.0.0.0        port: 7773        allow: 127.0.0.1        allow: 192.168replication:        binlog: yes        sync_speed: -1        slaveof:                type: sync                ip: 127.0.0.1                port: 7771logger:        level: error        output: /usr/data/ssdb_7773.log        rotate:                size: 1000000000leveldb:        cache_size: 500        block_size: 32        write_buffer_size: 64        compaction_speed: 1000        compression: yes

配置文件使用Tab而不是空格做缩排,(复制到配置文件后请把空格替换为Tab)。主从关系:7770(主)–>7772(从),7771(主)—>7773(从);配置文件如何配置请参考https://github.com/ideawu/ssdb-docs/blob/master/src/zh_cn/config.md。

创建工作目录

mkdir -p /usr/data/ssdb_7770mkdir -p /usr/data/ssdb_7771mkdir -p /usr/data/ssdb_7772mkdir -p /usr/data/ssdb_7773

启动

nohup /usr/servers/ssdb-1.8.0/ssdb-server  /usr/chapter7/ssdb_basic_7770.conf &nohup /usr/servers/ssdb-1.8.0/ssdb-server  /usr/chapter7/ssdb_basic_7771.conf &nohup /usr/servers/ssdb-1.8.0/ssdb-server  /usr/chapter7/ssdb_basic_7772.conf &nohup /usr/servers/ssdb-1.8.0/ssdb-server  /usr/chapter7/ssdb_basic_7773.conf &

通过ps -aux | grep ssdb命令看是否启动了,tail -f nohup.out查看错误信息。

商品介绍SSDB集群配置

vim /usr/chapter7/ssdb_desc_8880.conf

work_dir = /usr/data/ssdb_8880pidfile = /usr/data/ssdb8880.pidserver:        ip: 0.0.0.0        port: 8880        allow: 127.0.0.1        allow: 192.168replication:        binlog: yes        sync_speed: -1        slaveof:logger:        level: error        output: /usr/data/ssdb_8880.log        rotate:                size: 1000000000leveldb:        cache_size: 500        block_size: 32        write_buffer_size: 64        compaction_speed: 1000        compression: yes

vim /usr/chapter7/ssdb_desc_8881.conf

work_dir = /usr/data/ssdb_8881pidfile = /usr/data/ssdb8881.pidserver:        ip: 0.0.0.0        port: 8881        allow: 127.0.0.1        allow: 192.168logger:        level: error        output: /usr/data/ssdb_8881.log        rotate:                size: 1000000000leveldb:        cache_size: 500        block_size: 32        write_buffer_size: 64        compaction_speed: 1000        compression: yes

创建工作目录

mkdir -p /usr/data/ssdb_888{0,1,2,3}  

启动

nohup /usr/servers/ssdb-1.8.0/ssdb-server  /usr/chapter7/ssdb_desc_8880.conf &nohup /usr/servers/ssdb-1.8.0/ssdb-server  /usr/chapter7/ssdb_desc_8881.conf &nohup /usr/servers/ssdb-1.8.0/ssdb-server  /usr/chapter7/ssdb_desc_8882.conf &nohup /usr/servers/ssdb-1.8.0/ssdb-server  /usr/chapter7/ssdb_desc_8883.conf &

其他信息Redis配置

port 6660pidfile "/var/run/redis_6660.pid"#设置内存大小,根据实际情况设置,此处测试仅设置20mbmaxmemory 20mb#内存不足时,所有KEY按照LRU算法删除maxmemory-policy allkeys-lru#Redis的过期算法不是精确的而是通过采样来算的,默认采样为3个,此处我们改成10maxmemory-samples 10#不进行RDB持久化save “”#不进行AOF持久化appendonly no

如上配置放到配置文件最末尾即可;此处内存不足时的驱逐算法为所有KEY按照LRU进行删除(实际是内存基本上不会遇到满的情况);主从关系:6660(主)–>6661(从)和6660(主)–>6662(从)。

启动

nohup /usr/servers/redis-2.8.19/src/redis-server /usr/chapter7/redis_6660.conf &  nohup /usr/servers/redis-2.8.19/src/redis-server /usr/chapter7/redis_6661.conf &  nohup /usr/servers/redis-2.8.19/src/redis-server /usr/chapter7/redis_6662.conf & 

测试

测试时在主SSDB/Redis中写入数据,然后从从SSDB/Redis能读取到数据即表示配置主从成功。
测试商品基本信息SSDB集群

root@kaitao:/usr/chapter7# /usr/servers/redis-2.8.19/src/redis-cli  -p 7770127.0.0.1:7770> set i 1OK127.0.0.1:7770> root@kaitao:/usr/chapter7# /usr/servers/redis-2.8.19/src/redis-cli  -p 7772127.0.0.1:7772> get i"1"

Twemproxy配置

vim /usr/chapter7/nutcracker.yml

basic_master:  listen: 127.0.0.1:1111  hash: fnv1a_64  distribution: ketama  redis: true  timeout: 1000  hash_tag: "::"  servers:   - 127.0.0.1:7770:1 server1   - 127.0.0.1:7771:1 server2basic_slave:  listen: 127.0.0.1:1112  hash: fnv1a_64  distribution: ketama  redis: true  timeout: 1000  hash_tag: "::"  servers:   - 127.0.0.1:7772:1 server1   - 127.0.0.1:7773:1 server2desc_master:  listen: 127.0.0.1:1113  hash: fnv1a_64  distribution: ketama  redis: true  timeout: 1000  hash_tag: "::"  servers:   - 127.0.0.1:8880:1 server1   - 127.0.0.1:8881:1 server2desc_slave:  listen: 127.0.0.1:1114  hash: fnv1a_64  distribution: ketama  redis: true  timeout: 1000  servers:   - 127.0.0.1:8882:1 server1   - 127.0.0.1:8883:1 server2other_master:  listen: 127.0.0.1:1115  hash: fnv1a_64  distribution: random  redis: true  timeout: 1000  hash_tag: "::"  servers:   - 127.0.0.1:6660:1 server1other_slave:  listen: 127.0.0.1:1116  hash: fnv1a_64  distribution: random  redis: true  timeout: 1000  hash_tag: "::"  servers:   - 127.0.0.1:6661:1 server1   - 127.0.0.1:6662:1 server2

1、因为我们使用了主从,所以需要给server起一个名字如server1、server2;否则分片算法默认根据ip:port:weight,这样就会主从数据的分片算法不一致;

2、其他信息Redis因为每个Redis是对等的,因此分片算法可以使用random;

3、我们使用了hash_tag,可以保证相同的tag在一个分片上(本例配置了但没有用到该特性)。

复制第六章的nutcracker.init,帮把配置文件改为usr/chapter7/nutcracker.yml。然后通过/usr/chapter7/nutcracker.init start启动Twemproxy。
测试主从集群是否工作正常:

root@kaitao:/usr/chapter7# /usr/servers/redis-2.8.19/src/redis-cli -p 1111127.0.0.1:1111> set i 1OK127.0.0.1:1111> root@kaitao:/usr/chapter7# /usr/servers/redis-2.8.19/src/redis-cli -p 1112127.0.0.1:1112> get i"1"127.0.0.1:1112> root@kaitao:/usr/chapter7# /usr/servers/redis-2.8.19/src/redis-cli -p 1113127.0.0.1:1113> set i 1OK127.0.0.1:1113> root@kaitao:/usr/chapter7# /usr/servers/redis-2.8.19/src/redis-cli -p 1114127.0.0.1:1114> get i"1"127.0.0.1:1114> root@kaitao:/usr/chapter7# /usr/servers/redis-2.8.19/src/redis-cli -p 1115127.0.0.1:1115> set i 1OK127.0.0.1:1115> root@kaitao:/usr/chapter7# /usr/servers/redis-2.8.19/src/redis-cli -p 1116127.0.0.1:1116> get i"1"

动态服务实现

因为真实数据是从多个子系统获取,很难模拟这么多子系统交互,所以此处我们使用假数据来进行实现。

项目搭建

我们使用Maven搭建Web项目,Maven知识请自行学习。

项目依赖

本文将最小化依赖,即仅依赖我们需要的servlet、jackson、guava、jedis。

  <dependencies>    <dependency>      <groupId>javax.servlet</groupId>      <artifactId>javax.servlet-api</artifactId>      <version>3.0.1</version>      <scope>provided</scope>    </dependency>    <dependency>      <groupId>com.google.guava</groupId>      <artifactId>guava</artifactId>      <version>17.0</version>    </dependency>    <dependency>      <groupId>redis.clients</groupId>      <artifactId>jedis</artifactId>      <version>2.5.2</version>    </dependency>    <dependency>      <groupId>com.fasterxml.jackson.core</groupId>      <artifactId>jackson-core</artifactId>      <version>2.3.3</version>    </dependency>    <dependency>      <groupId>com.fasterxml.jackson.core</groupId>      <artifactId>jackson-databind</artifactId>      <version>2.3.3</version>    </dependency>  </dependencies>

guava是类似于apache commons的一个基础类库,用于简化一些重复操作,可以参考http://ifeve.com/google-guava/。

核心代码

com.github.zhangkaitao.chapter7.servlet.ProductServiceServlet

    @Override    protected void doGet(HttpServletRequest req, HttpServletResponse resp) throws ServletException, IOException {        String type = req.getParameter("type");        String content = null;        try {            if("basic".equals(type)) {                content = getBasicInfo(req.getParameter("skuId"));            } else if("desc".equals(type)) {                content = getDescInfo(req.getParameter("skuId"));            } else if("other".equals(type)) {                content = getOtherInfo(req.getParameter("ps3Id"), req.getParameter("brandId"));            }        } catch (Exception e) {            e.printStackTrace();            resp.setStatus(HttpServletResponse.SC_INTERNAL_SERVER_ERROR);            return;        }        if(content != null) {            resp.setCharacterEncoding("UTF-8");            resp.getWriter().write(content);        } else {            resp.setStatus(HttpServletResponse.SC_NOT_FOUND);        }    }

根据请求参数type来决定调用哪个服务获取数据。

基本信息服务

   private String getBasicInfo(String skuId) throws Exception {        Map<String, Object> map = new HashMap<String, Object>();        //商品编号        map.put("skuId", skuId);        //名称        map.put("name", "苹果(Apple)iPhone 6 (A1586) 16GB 金色 移动联通电信4G手机");        //一级二级三级分类        map.put("ps1Id", 9987);        map.put("ps2Id", 653);        map.put("ps3Id", 655);        //品牌ID        map.put("brandId", 14026);        //图片列表        map.put("imgs", getImgs(skuId));        //上架时间        map.put("date", "2014-10-09 22:29:09");        //商品毛重        map.put("weight", "400");        //颜色尺码        map.put("colorSize", getColorSize(skuId));        //扩展属性        map.put("expands", getExpands(skuId));        //规格参数        map.put("propCodes", getPropCodes(skuId));        map.put("date", System.currentTimeMillis());        String content = objectMapper.writeValueAsString(map);        //实际应用应该是发送MQ        asyncSetToRedis(basicInfoJedisPool, "p:" + skuId + ":", content);        return objectMapper.writeValueAsString(map);    }    private List<String> getImgs(String skuId) {        return Lists.newArrayList(                "jfs/t277/193/1005339798/768456/29136988/542d0798N19d42ce3.jpg",                "jfs/t352/148/1022071312/209475/53b8cd7f/542d079bN3ea45c98.jpg",                "jfs/t274/315/1008507116/108039/f70cb380/542d0799Na03319e6.jpg",                "jfs/t337/181/1064215916/27801/b5026705/542d079aNf184ce18.jpg"        );    }    private List<Map<String, Object>> getColorSize(String skuId) {        return Lists.newArrayList(            makeColorSize(1217499, "金色", "公开版(16GB ROM)"),            makeColorSize(1217500, "深空灰", "公开版(16GB ROM)"),            makeColorSize(1217501, "银色", "公开版(16GB ROM)"),            makeColorSize(1217508, "金色", "公开版(64GB ROM)"),            makeColorSize(1217509, "深空灰", "公开版(64GB ROM)"),            makeColorSize(1217509, "银色", "公开版(64GB ROM)"),            makeColorSize(1217493, "金色", "移动4G版 (16GB)"),            makeColorSize(1217494, "深空灰", "移动4G版 (16GB)"),            makeColorSize(1217495, "银色", "移动4G版 (16GB)"),            makeColorSize(1217503, "金色", "移动4G版 (64GB)"),            makeColorSize(1217503, "金色", "移动4G版 (64GB)"),            makeColorSize(1217504, "深空灰", "移动4G版 (64GB)"),            makeColorSize(1217505, "银色", "移动4G版 (64GB)")        );    }    private Map<String, Object> makeColorSize(long skuId, String color, String size) {        Map<String, Object> cs1 = Maps.newHashMap();        cs1.put("SkuId", skuId);        cs1.put("Color", color);        cs1.put("Size", size);        return cs1;    }    private List<List<?>> getExpands(String skuId) {        return Lists.newArrayList(                (List<?>)Lists.newArrayList("热点", Lists.newArrayList("超薄7mm以下", "支持NFC")),                (List<?>)Lists.newArrayList("系统", "苹果(IOS)"),                (List<?>)Lists.newArrayList("系统", "苹果(IOS)"),                (List<?>)Lists.newArrayList("购买方式", "非合约机")        );    }    private Map<String, List<List<String>>> getPropCodes(String skuId) {        Map<String, List<List<String>>> map = Maps.newHashMap();        map.put("主体", Lists.<List<String>>newArrayList(                Lists.<String>newArrayList("品牌", "苹果(Apple)"),                Lists.<String>newArrayList("型号", "iPhone 6 A1586"),                Lists.<String>newArrayList("颜色", "金色"),                Lists.<String>newArrayList("上市年份", "2014年")        ));        map.put("存储", Lists.<List<String>>newArrayList(                Lists.<String>newArrayList("机身内存", "16GB ROM"),                Lists.<String>newArrayList("储存卡类型", "不支持")        ));        map.put("显示", Lists.<List<String>>newArrayList(                Lists.<String>newArrayList("屏幕尺寸", "4.7英寸"),                Lists.<String>newArrayList("触摸屏", "Retina HD"),                Lists.<String>newArrayList("分辨率", "1334 x 750")        ));        return map;    }
    private ObjectMapper objectMapper = new ObjectMapper();    private JedisPool basicInfoJedisPool = createJedisPool("127.0.0.1", 1111);    private JedisPool descInfoJedisPool = createJedisPool("127.0.0.1", 1113);    private JedisPool otherInfoJedisPool = createJedisPool("127.0.0.1", 1115);    private JedisPool createJedisPool(String host, int port) {        GenericObjectPoolConfig poolConfig = new GenericObjectPoolConfig();        poolConfig.setMaxTotal(100);        return new JedisPool(poolConfig, host, port);    }    private ExecutorService executorService = Executors.newFixedThreadPool(10);    private void asyncSetToRedis(final JedisPool jedisPool, final String key, final String content) {        executorService.submit(new Runnable() {            @Override            public void run() {                Jedis jedis = null;                try {                    jedis = jedisPool.getResource();                    jedis.set(key, content);                } catch (Exception e) {                    e.printStackTrace();                    jedisPool.returnBrokenResource(jedis);                } finally {                    jedisPool.returnResource(jedis);                }            }        });    }

本例使用Jackson进行JSON的序列化;Jedis进行Redis的操作;使用线程池做异步更新(实际应用中可以使用MQ做实现)。

nginx配置

vim /usr/chapter7/nginx_chapter7.conf

upstream backend {    server 127.0.0.1:8080 max_fails=5 fail_timeout=10s weight=1;    check interval=3000 rise=1 fall=2 timeout=5000 type=tcp default_down=false;    keepalive 100;}server {    listen       80;    server_name  _    location ~ /backend/(.*) {        #internal;        keepalive_timeout   30s;        keepalive_requests  1000;        #支持keep-alive        proxy_http_version 1.1;        proxy_set_header Connection "";        rewrite /backend(/.*) $1 break;        proxy_pass_request_headers off;        #more_clear_input_headers Accept-Encoding;        proxy_next_upstream error timeout;        proxy_pass http://backend;    }}

实际生产环境要把#internal打开,表示只有本nginx能访问。

vim /usr/servers/nginx/conf/nginx.conf

include /usr/chapter7/nginx_chapter7.conf;#为了方便测试,注释掉example.confinclude /usr/chapter6/nginx_chapter6.conf;
    #lua模块路径,其中”;;”表示默认搜索路径,默认到/usr/servers/nginx下找      lua_package_path "/usr/chapter7/lualib/?.lua;;";  #lua 模块      lua_package_cpath "/usr/chapter7/lualib/?.so;;";  #c模块  

lua模块从/usr/chapter7目录加载,因为我们要写自己的模块使用。

重启nginx

/usr/servers/nginx/sbin/nginx -s reload

前端展示实现

我们分为三部分实现:基础组件、商品介绍、前端展示部分。

基础组件

首先我们进行基础组件的实现,商品介绍和前端展示部分都需要读取Redis和Http服务,因此我们可以抽取公共部分出来复用。

vim /usr/chapter7/lualib/item/common.lualocal redis = require("resty.redis")local ngx_log = ngx.loglocal ngx_ERR = ngx.ERRlocal function close_redis(red)    if not red then        return    end    --释放连接(连接池实现)    local pool_max_idle_time = 10000 --毫秒    local pool_size = 100 --连接池大小    local ok, err = red:set_keepalive(pool_max_idle_time, pool_size)    if not ok then        ngx_log(ngx_ERR, "set redis keepalive error : ", err)    endendlocal function read_redis(ip, port, keys)    local red = redis:new()    red:set_timeout(1000)    local ok, err = red:connect(ip, port)    if not ok then        ngx_log(ngx_ERR, "connect to redis error : ", err)        return close_redis(red)    end    local resp = nil    if #keys == 1 then        resp, err = red:get(keys[1])    else        resp, err = red:mget(keys)    end    if not resp then        ngx_log(ngx_ERR, "get redis content error : ", err)        return close_redis(red)    end    --得到的数据为空处理    if resp == ngx.null then        resp = nil    end    close_redis(red)    return respendlocal function read_http(args)    local resp = ngx.location.capture("/backend/info", {        method = ngx.HTTP_GET,        args = args    })    if not resp then        ngx_log(ngx_ERR, "request error")        return    end    if resp.status ~= 200 then        ngx_log(ngx_ERR, "request error, status :", resp.status)        return    end    return resp.bodyendlocal _M = {    read_redis = read_redis,    read_http = read_http}return _M

整个逻辑和第六章类似;只是read_redis根据参数keys个数支持get和mget。 比如read_redis(ip, port, {“key1”})则调用get而read_redis(ip, port, {“key1”, “key2”})则调用mget。

商品介绍

核心代码

vim /usr/chapter7/desc.lualocal common = require("item.common")local read_redis = common.read_redislocal read_http = common.read_httplocal ngx_log = ngx.loglocal ngx_ERR = ngx.ERRlocal ngx_exit = ngx.exitlocal ngx_print = ngx.printlocal ngx_re_match = ngx.re.matchlocal ngx_var = ngx.varlocal descKey = "d:" .. skuId .. ":"local descInfoStr = read_redis("127.0.0.1", 1114, {descKey})if not descInfoStr then   ngx_log(ngx_ERR, "redis not found desc info, back to http, skuId : ", skuId)   descInfoStr = read_http({type="desc", skuId = skuId})endif not descInfoStr then   ngx_log(ngx_ERR, "http not found basic info, skuId : ", skuId)   return ngx_exit(404)endngx_print("showdesc(")ngx_print(descInfoStr)ngx_print(")")
ngx_print(")")  

通过复用逻辑后整体代码简化了许多;此处读取商品介绍从集群;另外前端展示使用JSONP技术展示商品介绍。

nginx配置

vim /usr/chapter7/nginx_chapter7.conf

    location ~^/desc/(\d+)$ {        if ($host != "d.3.cn") {           return 403;        }        default_type application/x-javascript;        charset utf-8;        lua_code_cache on;        set $skuId $1;        content_by_lua_file /usr/chapter7/desc.lua;    }

前端展示

核心代码

vim /usr/chapter7/item.lua

local common = require("item.common")local item = require("item")local read_redis = common.read_redislocal read_http = common.read_httplocal cjson = require("cjson")local cjson_decode = cjson.decodelocal ngx_log = ngx.loglocal ngx_ERR = ngx.ERRlocal ngx_exit = ngx.exitlocal ngx_print = ngx.printlocal ngx_var = ngx.varlocal skuId = ngx_var.skuId--获取基本信息local basicInfoKey = "p:" .. skuId .. ":"local basicInfoStr = read_redis("127.0.0.1", 1112, {basicInfoKey})if not basicInfoStr then   ngx_log(ngx_ERR, "redis not found basic info, back to http, skuId : ", skuId)   basicInfoStr = read_http({type="basic", skuId = skuId})endif not basicInfoStr then   ngx_log(ngx_ERR, "http not found basic info, skuId : ", skuId)   return ngx_exit(404)endlocal basicInfo = cjson_decode(basicInfoStr)local ps3Id = basicInfo["ps3Id"]local brandId = basicInfo["brandId"]--获取其他信息local breadcrumbKey = "s:" .. ps3Id .. ":"local brandKey = "b:" .. brandId ..":"local otherInfo = read_redis("127.0.0.1", 1116, {breadcrumbKey, brandKey}) or {}local breadcrumbStr = otherInfo[1]local brandStr = otherInfo[2]if breadcrumbStr then   basicInfo["breadcrumb"] = cjson_decode(breadcrumbStr)endif brandStr then   basicInfo["brand"] = cjson_decode(brandStr)endif not breadcrumbStr and not brandStr then   ngx_log(ngx_ERR, "redis not found other info, back to http, skuId : ", brandId)   local otherInfoStr = read_http({type="other", ps3Id = ps3Id, brandId = brandId})   if not otherInfoStr then       ngx_log(ngx_ERR, "http not found other info, skuId : ", skuId)   else     local otherInfo = cjson_decode(otherInfoStr)     basicInfo["breadcrumb"] = otherInfo["breadcrumb"]     basicInfo["brand"] = otherInfo["brand"]   endendlocal name = basicInfo["name"]--name to unicodebasicInfo["unicodeName"] = item.utf8_to_unicode(name)--字符串截取,超长显示...basicInfo["moreName"] = item.trunc(name, 10)--初始化各分类的urlitem.init_breadcrumb(basicInfo)--初始化扩展属性item.init_expand(basicInfo)--初始化颜色尺码item.init_color_size(basicInfo)local template = require "resty.template"template.caching(true)template.render("item.html", basicInfo)

整个逻辑分为四部分:1、获取基本信息;2、根据基本信息中的关联关系获取其他信息;3、初始化/格式化数据;4、渲染模板。

初始化模块

vim /usr/chapter7/lualib/item.lua

local bit = require("bit")local utf8 = require("utf8")local cjson = require("cjson")local cjson_encode = cjson.encodelocal bit_band = bit.bandlocal bit_bor = bit.borlocal bit_lshift = bit.lshiftlocal string_format = string.formatlocal string_byte = string.bytelocal table_concat = table.concat--utf8转为unicodelocal function utf8_to_unicode(str)    if not str or str == "" or str == ngx.null then        return nil    end    local res, seq, val = {}, 0, nil    for i = 1, #str do        local c = string_byte(str, i)        if seq == 0 then            if val then                res[#res + 1] = string_format("%04x", val)            end           seq = c < 0x80 and 1 or c < 0xE0 and 2 or c < 0xF0 and 3 or                              c < 0xF8 and 4 or --c < 0xFC and 5 or c < 0xFE and 6 or                              0            if seq == 0 then                ngx.log(ngx.ERR, 'invalid UTF-8 character sequence' .. ",,," .. tostring(str))                return str            end            val = bit_band(c, 2 ^ (8 - seq) - 1)        else            val = bit_bor(bit_lshift(val, 6), bit_band(c, 0x3F))        end        seq = seq - 1    end    if val then        res[#res + 1] = string_format("%04x", val)    end    if #res == 0 then        return str    end    return "\\u" .. table_concat(res, "\\u")end--utf8字符串截取local function trunc(str, len)   if not str then     return nil   end   if utf8.len(str) > len then      return utf8.sub(str, 1, len) .. "..."   end   return strend--初始化面包屑local function init_breadcrumb(info)    local breadcrumb = info["breadcrumb"]    if not breadcrumb then       return    end    local ps1Id = breadcrumb[1][1]    local ps2Id = breadcrumb[2][1]    local ps3Id = breadcrumb[3][1]    --此处应该根据一级分类查找url    local ps1Url = "http://shouji.jd.com/"    local ps2Url = "http://channel.jd.com/shouji.html"    local ps3Url = "http://list.jd.com/list.html?cat=" .. ps1Id .. "," .. ps2Id .. "," .. ps3Id    breadcrumb[1][3] = ps1Url    breadcrumb[2][3] = ps2Url    breadcrumb[3][3] = ps3Urlend--初始化扩展属性local function init_expand(info)   local expands = info["expands"]   if not expands then     return   end   for _, e in ipairs(expands) do      if type(e[2]) == "table" then         e[2] = table_concat(e[2], ",")      end   endend--初始化颜色尺码local function init_color_size(info)   local colorSize = info["colorSize"]   --颜色尺码JSON串   local colorSizeJson = cjson_encode(colorSize)   --颜色列表(不重复)   local colorList = {}   --尺码列表(不重复)   local sizeList = {}   info["colorSizeJson"] = colorSizeJson   info["colorList"] = colorList   info["sizeList"] = sizeList   local colorSet = {}   local sizeSet = {}   for _, cz in ipairs(colorSize) do      local color = cz["Color"]      local size = cz["Size"]      if color and color ~= "" and not colorSet[color] then         colorList[#colorList + 1] = {color = color, url = "http://item.jd.com/" ..cz["SkuId"] .. ".html"}         colorSet[color] = true      end      if size and size ~= "" and not sizeSet[size] then         sizeList[#sizeList + 1] = {size = size, url = "http://item.jd.com/" ..cz["SkuId"] .. ".html"}         sizeSet[size] = ""      end   endendlocal _M = {   utf8_to_unicode = utf8_to_unicode,   trunc = trunc,   init_breadcrumb = init_breadcrumb,   init_expand = init_expand,   init_color_size = init_color_size}return _M

模板html片段

       var pageConfig = {            compatible: true,            product: {                skuid: {* skuId *},                name: '{* unicodeName *}',                skuidkey:'AFC266E971535B664FC926D34E91C879',                href: 'http://item.jd.com/{* skuId *}.html',                src: '{* imgs[1] *}',                cat: [{* ps1Id *},{* ps2Id *},{* ps3Id *}],                brand: {* brandId *},                tips: false,                pType: 1,                venderId:0,                shopId:'0',                specialAttrs:["HYKHSP-0","isDistribution","isHaveYB","isSelfService-0","isWeChatStock-0","packType","IsNewGoods","isCanUseDQ","isSupportCard","isCanUseJQ","isOverseaPurchase-0","is7ToReturn-1","isCanVAT"],                videoPath:'',                desc: 'http://d.3.cn/desc/{* skuId *}'            }        };        var warestatus = 1;                        {% if colorSizeJson then %} var ColorSize = {* colorSizeJson *};{% end %}                {-raw-}                try{(function(flag){ if(!flag){return;} if(window.location.hash == '#m'){var exp = new Date();exp.setTime(exp.getTime() + 30 * 24 * 60 * 60 * 1000);document.cookie = "pcm=1;expires=" + exp.toGMTString() + ";path=/;domain=jd.com";return;}else{var cook=document.cookie.match(new RegExp("(^| )pcm=([^;]*)(;|$)"));if(cook&&cook.length>2&&unescape(cook[2])=="2"){flag=false;}} var userAgent = navigator.userAgent; if(userAgent){ userAgent = userAgent.toUpperCase();if(userAgent.indexOf("PAD")>-1){return;} var mobilePhoneList = ["IOS","IPHONE","ANDROID","WINDOWS PHONE"];for(var i=0,len=mobilePhoneList.length;i<len;i++){ if(userAgent.indexOf(mobilePhoneList[i])>-1){var url="http://m.jd.com/product/"+pageConfig.product.skuid+".html";if(flag){window.showtouchurl=true;}else{window.location.href = url;}break;}}}})((function(){var json={"6881":3,"1195":3,"10011":3,"6980":3,"12360":3};if(json[pageConfig.product.cat[0]+""]==1||json[pageConfig.product.cat[1]+""]==2||json[pageConfig.product.cat[2]+""]==3){return false;}else{return true;}})());}catch(e){}                {-raw-}

面包屑

<div class="breadcrumb">    <strong><a href='{* breadcrumb[1][3] *}'>{* breadcrumb[1][2] *}</a></strong>    <span>        &nbsp;&gt;&nbsp;        <a href='{* breadcrumb[2][3] *}'>{* breadcrumb[2][2] *}</a>        &nbsp;&gt;&nbsp;        <a href='{* breadcrumb[3][3] *}'>{* breadcrumb[3][2] *}</a>        &nbsp;&gt;&nbsp;    </span>    <span>        {% if brand then %}        <a href='http://www.jd.com/pinpai/{* ps3Id *}-{* brandId *}.html'>{* brand['name'] *}</a>        &nbsp;&gt;&nbsp;       {% end %}       <a href='http://item.jd.com/{* skuId *}.html'>{* moreName *}</a>    </span></div>

nginx配置

    #模板加载位置    set $template_root "/usr/chapter7";    location ~ ^/(\d+).html$ {        default_type 'text/html';        charset utf-8;        lua_code_cache on;        set $skuId $1;        content_by_lua_file /usr/chapter7/item.lua;    }

优化

local cache

对于其他信息,对数据一致性要求不敏感,而且数据量很少,完全可以在本地缓存全量;而且可以设置如5-10分钟的过期时间是完全可以接受的;因此可以lua_shared_dict全局内存进行缓存。具体逻辑可以参考

local nginx_shared = ngx.shared--item.jd.com配置的缓存local local_cache = nginx_shared.item_local_cachelocal function cache_get(key)    if not local_cache then        return nil    end    return local_cache:get(key)endlocal function cache_set(key, value)    if not local_cache then        return nil    end    return local_cache:set(key, value, 10 * 60) --10分钟endlocal function get(ip, port, keys)    local tables = {}    local fetchKeys = {}    local resp = nil    local status = STATUS_OK    --如果tables是个map #tables拿不到长度    local has_value = false    --先读取本地缓存    for i, key in ipairs(keys) do        local value = cache_get(key)        if value then            if value == "" then                value = nil            end            tables[key] = value            has_value = true        else            fetchKeys[#fetchKeys + 1] = key        end    end    --如果还有数据没获取 从redis获取    if #fetchKeys > 0 then        if #fetchKeys == 1 then            status, resp = redis_get(ip, port, fetchKeys[1])        else            status, resp = redis_mget(ip, port, fetchKeys)        end        if status == STATUS_OK then            for i = 1, #fetchKeys do                 local key = fetchKeys[i]                 local value = nil                 if #fetchKeys == 1 then                    value = resp                 else                    value = get_data(resp, i)                 end                 tables[key] = value                  has_value = true                  cache_set(key, value or "", ttl)            end        end    end    --如果从缓存查到 就认为ok    if has_value and status == STATUS_NOT_FOUND then        status = STATUS_OK    end    return status, tablesend

nginx proxy cache

为了防止恶意刷页面/热点页面访问频繁,我们可以使用nginx proxy_cache做页面缓存,当然更好的选择是使用CDN技术,如通过Apache Traffic Server、Squid、Varnish。

1、nginx.conf配置

    proxy_buffering on;    proxy_buffer_size 8k;    proxy_buffers 256 8k;    proxy_busy_buffers_size 64k;    proxy_temp_file_write_size 64k;    proxy_temp_path /usr/servers/nginx/proxy_temp;    #设置Web缓存区名称为cache_one,内存缓存空间大小为200MB,1分钟没有被访问的内容自动清除,硬盘缓存空间大小为30GB。    proxy_cache_path  /usr/servers/nginx/proxy_cache levels=1:2 keys_zone=cache_item:200m inactive=1m max_size=30g;

增加proxy_cache的配置,可以通过挂载一块内存作为缓存的存储空间。更多配置规则请参考http://nginx.org/cn/docs/http/ngx_http_proxy_module.html。

2、nginx_chapter7.conf配置

与server指令配置同级

############ 测试时使用的动态请求map $host $item_dynamic {    default                    "0";    item2015.jd.com            "1";}

即如果域名为item2015.jd.com则item_dynamic=1。

   location ~ ^/(\d+).html$ {        set $skuId $1;        if ($host !~ "^(item|item2015)\.jd\.com$") {           return 403;        }        expires 3m;        proxy_cache cache_item;        proxy_cache_key $uri;        proxy_cache_bypass $item_dynamic;        proxy_no_cache $item_dynamic;        proxy_cache_valid 200 301 3m;        proxy_cache_use_stale updating error timeout invalid_header http_500 http_502 http_503 http_504;        proxy_pass_request_headers off;        proxy_set_header Host $host;        #支持keep-alive        proxy_http_version 1.1;        proxy_set_header Connection "";        proxy_pass http://127.0.0.1/proxy/$skuId.html;        add_header X-Cache '$upstream_cache_status';    }    location ~ ^/proxy/(\d+).html$ {        allow 127.0.0.1;        deny all;        keepalive_timeout   30s;        keepalive_requests  1000;        default_type 'text/html';        charset utf-8;        lua_code_cache on;        set $skuId $1;        content_by_lua_file /usr/chapter7/item.lua;    }

expires:设置响应缓存头信息,此处是3分钟;将会得到Cache-Control:max-age=180和类似Expires:Sat, 28 Feb 2015 10:01:10 GMT的响应头;
proxy_cache:使用之前在nginx.conf中配置的cache_item缓存;

proxy_cache_key:缓存key为uri,不包括host和参数,这样不管用户怎么通过在url上加随机数都是走缓存的;

proxy_cache_bypass:nginx不从缓存取响应的条件,可以写多个;如果存在一个字符串条件且不是“0”,那么nginx就不会从缓存中取响应内容;此处如果我们使用的host为item2015.jd.com时就不会从缓存取响应内容;

proxy_no_cache:nginx不将响应内容写入缓存的条件,可以写多个;如果存在一个字符串条件且不是“0”,那么nginx就不会从将响应内容写入缓存;此处如果我们使用的host为item2015.jd.com时就不会将响应内容写入缓存;

proxy_cache_valid:为不同的响应状态码设置不同的缓存时间,此处我们对200、301缓存3分钟;

proxy_cache_use_stale:什么情况下使用不新鲜(过期)的缓存内容;配置和proxy_next_upstream内容类似;此处配置了如果连接出错、超时、404、500等都会使用不新鲜的缓存内容;此外我们配置了updating配置,通过配置它可以在nginx正在更新缓存(其中一个Worker进程)时(其他的Worker进程)使用不新鲜的缓存进行响应,这样可以减少回源的数量;

proxy_pass_request_headers:我们不需要请求头,所以不传递;

proxy_http_version 1.1和proxy_set_header Connection “”:支持keepalive;

add_header X-Cache ‘$upstream_cache_status’:添加是否缓存命中的响应头;比如命中HIT、不命中MISS、不走缓存BYPASS;比如命中会看到X-Cache:HIT响应头;

allow/deny:允许和拒绝访问的ip列表,此处我们只允许本机访问;

keepalive_timeout 30s和keepalive_requests 1000:支持keepalive;

nginx_chapter7.conf清理缓存配置

    location /purge {        allow     127.0.0.1;        allow     192.168.0.0/16;        deny      all;        proxy_cache_purge  cache_item $arg_url;    }

3、修改item.lua代码

--添加Last-Modified,用于响应304缓存ngx.header["Last-Modified"] = ngx.http_time(ngx.now())local template = require "resty.template"template.caching(true)template.render("item.html", basicInfo)~                                         

在渲染模板前设置Last-Modified,用于判断内容是否变更的条件,默认Nginx通过等于去比较,也可以通过配置if_modified_since指令来支持小于等于比较;如果请求头发送的If-Modified-Since和Last-Modified匹配则返回304响应,即内容没有变更,使用本地缓存。此处可能看到了我们的Last-Modified是当前时间,不是商品信息变更的时间;商品信息变更时间由:商品信息变更时间、面包屑变更时间和品牌变更时间三者决定的,因此实际应用时应该取三者最大的;还一个问题就是模板内容可能变了,但是商品信息没有变,此时使用Last-Modified得到的内容可能是错误的,所以可以通过使用ETag技术来解决这个问题,ETag可以认为是内容的一个摘要,内容变更后摘要就变了。

GZIP压缩

修改nginx.conf配置文件

  gzip on;   gzip_min_length  4k;   gzip_buffers     4 16k;   gzip_http_version 1.0;   gzip_proxied        any;  #前端是squid的情况下要加此参数,否则squid上不缓存gzip文件   gzip_comp_level 2;   gzip_types       text/plain application/x-javascript text/css application/xml;   gzip_vary on;

此处我们指定至少4k时才压缩,如果数据太小压缩没有意义。

0 0
原创粉丝点击