MongoDB（online） 优化

MongoDB（online） 优化

• 1. find、findOne
• 2. 操作 vip_emp_relation 的一个公共方法
• 3. 查询记录数
• 4. save、insert
• 5. 总结

1. find、findOne

• 项目实例

MongoTemplate mongoTemplate = mongodbClient.getMongoTemplate();
DBCollection cursor = mongoTemplate.getCollection("vip_batchsend_message");
BasicDBObject query = new BasicDBObject();
query.put("sms_code", sms_code);
query.put("open_id", open_id);

JSONObject message = new JSONObject();

DBCursor cursor1 = cursor.find(query);
LOGGER.info("-------------sms_code:" + sms_code + "----open_id:" + open_id + "--" + cursor1.size());
if (cursor1.size() > 0) {
    DBObject dBObject = cursor1.next();
    String corp_code = dBObject.get("corp_code").toString();
    .....

• 简单阐述
  1. 代码意图是如果记录存在就修改
  2. 只需要判断是否存在就完成目的
  3. find返回的是DBCursor，这里不合适，我们只需要知道是否存在即可
  4. 修改建议
     1. 固定的条件提前预先组合
     2. find改成findOne
     3. 判断只需要查询一条记录即可，用findOne就可以，直接获取一个Object，判断后即可根据Key获取Value进行后续操作
• 建议修改

BasicDBObject query = new BasicDBObject();
query.put("sms_code", sms_code);
query.put("open_id", open_id);
MongoTemplate mongoTemplate = mongodbClient.getMongoTemplate();
JSONObject message = new JSONObject();
DBCollection cursor = mongoTemplate.getCollection("vip_batchsend_message");
DBObject dbObject = cursor.findOne(query);
if (dbObject.isPartialObject()) {
    LOGGER.info("-------------sms_code:" + sms_code + "----open_id:" + open_id + "--" + 1);
    String corp_code = dbObject.get("corp_code").toString();
    String vip_id = dbObject.get("vip_id").toString();
    Data data_corp_code = new Data("corp_code", corp_code, ValueType.PARAM);
    Data data_vip_id = new Data("vip_ids", vip_id, ValueType.PARAM);
    ......

2. 操作 vip_emp_relation 的一个公共方法

• 项目实例

 public DBCursor selectRelation(String app_user_name, String open_id) throws SQLException {
        MongoTemplate mongoTemplate = mongodbClient.getMongoTemplate();
        DBCollection cursor = mongoTemplate.getCollection(WxConst.table_vip_emp_relation);

        Map keyMap = new HashMap();
        keyMap.put("_id", app_user_name + open_id);
        BasicDBObject queryCondition = new BasicDBObject();
        queryCondition.putAll(keyMap);
        DBCursor dbCursor = cursor.find(queryCondition);
        return dbCursor;
    }

• 简单阐述

  1. 查看此方法的引用，基本是做判断使用，没必要返回一个Cursor
  2. 如果引用的方法，有长任务，然后再操作，就会等待很长时间不释放资源

• 建议修改

   public DBObject selectRelation(String app_user_name, String open_id) throws SQLException {
      Map keyMap = new HashMap();
      keyMap.put("_id", app_user_name + open_id);
      BasicDBObject queryCondition = new BasicDBObject();
      queryCondition.putAll(keyMap);
      MongoTemplate mongoTemplate = mongodbClient.getMongoTemplate();
      DBCollection cursor = mongoTemplate.getCollection(WxConst.table_vip_emp_relation);
      DBObject dbCursor = cursor.findOne(queryCondition);
      return dbCursor;
  }

3. 查询记录数

• 项目实例

BasicDBObject basicDBObject=new BasicDBObject();
basicDBObject.put("activity_code",activity_code);
basicDBObject.put("open_id",open_id);
basicDBObject.put("status","1");
int count=cursor.find(basicDBObject).count();
basicDBObject.put("sign_status","Y");
int sign_count=cursor.find(basicDBObject).count();
if(sign_count>0){
    result="您已签到成功,请勿重复签到";
}else {
    if (count > 0) {
        BasicDBObject query=new BasicDBObject();
        query.put("activity_code",activity_code);
        query.put("open_id",open_id);
    ......

• 简单阐述

  1. 根据条件获取记录数
  2. 没必要先获取文档游标再查询记录数

• 建议修改

BasicDBObject basicDBObject=new BasicDBObject(); basicDBObject.put("activity_code",activity_code);
basicDBObject.put("open_id",open_id);
basicDBObject.put("status","1");
Long count=cursor.count(basicDBObject);
basicDBObject.put("sign_status","Y");
Long sign_count=cursor.count(basicDBObject);
if(sign_count>0){
    result="您已签到成功,请勿重复签到";
}else {
    if (count > 0) {
        BasicDBObject query=new BasicDBObject();
        query.put("activity_code",activity_code);
        query.put("open_id",open_id);
    ......

4. save、insert

• 项目实例

if (cursor.find(basicDBObject).count() > 0){
    BasicDBObject basicDBObject1=new BasicDBObject();
    basicDBObject1.put("sign_status","Y");
    basicDBObject1.put("sign_date",Common.DATETIME_FORMAT.format(new Date()));
    BasicDBObject update=new BasicDBObject();
    update.put("$set",basicDBObject1);
    cursor.update(basicDBObj,update,true,false);
}else {
    BasicDBObject dbObject = new BasicDBObject();
    dbObject.put("_id", app_id + "_" + activity_code  + "_" + open_id);
    dbObject.put("corp_code", corp_code);
    dbObject.put("sign_status","Y");
    dbObject.put("sign_date",Common.DATETIME_FORMAT.format(new Date()));
    dbObject.put("app_id", app_id);
    dbObject.put("activity_code", activity_code);
    dbObject.put("status", "0");
    dbObject.put("open_id", open_id);
    dbObject.put("vip", vip_array.getJSONObject(0));
    dbObject.put("modified_date", Common.DATETIME_FORMAT.format(new Date()));
    dbObject.put("created_date", Common.DATETIME_FORMAT.format(new Date()));
    cursor.save(dbObject);
}
    ......

• 简单阐述

  1. save 是先根据_id查询再修改，如果已经确认记录不存在可以省去查找的功能直接insert

• 建议修改

if (cursor.find(basicDBObject).count() > 0){
    BasicDBObject basicDBObject1=new BasicDBObject();
    basicDBObject1.put("sign_status","Y");
    basicDBObject1.put("sign_date",Common.DATETIME_FORMAT.format(new Date()));
    BasicDBObject update=new BasicDBObject();
    update.put("$set",basicDBObject1);
    cursor.update(basicDBObj,update,true,false);
}else {
    BasicDBObject dbObject = new BasicDBObject();
    dbObject.put("_id", app_id + "_" + activity_code  + "_" + open_id);
    dbObject.put("corp_code", corp_code);
    dbObject.put("sign_status","Y");
    dbObject.put("sign_date",Common.DATETIME_FORMAT.format(new Date()));
    dbObject.put("app_id", app_id);
    dbObject.put("activity_code", activity_code);
    dbObject.put("status", "0");
    dbObject.put("open_id", open_id);
    dbObject.put("vip", vip_array.getJSONObject(0));
    dbObject.put("modified_date", Common.DATETIME_FORMAT.format(new Date()));
    dbObject.put("created_date", Common.DATETIME_FORMAT.format(new Date()));
    cursor.insert(dbObject);
}
    ......

5. 总结

1. 以上是目前发现的问题及建议，会继续Review
2. 涉及到IO的操作标配就是尽晚打开尽早释放
3. 阿里的MongoDB默认没有启动读写分离（我已经测试确认过），我测试后会加上
4. 业务实现过程，尽可能结构化数据，介绍拼接出错或Key不存在的异常
5. 涉及到Cursor尽快手动关闭
6. 其它项目也可以参考，或者抛出来