sonic中syncd容器与redis容器通信源码解析

摘要

整个功能实现基本由以下步骤组成

1.syncd初始化配置

2.连接数据库并监听

3.数据处理

背景

sonic-buildimage的git commit为774778，将在此版本上进行分析，github地址如下：

https://github.com/Azure/sonic-buildimage/tree/77477857b47b114fde18afc33985e1a76c464c09

进入src目录，可以看到对应的代码如下图：

用到的代码为sairedis（在syncd上容器运行，负责与redis数据库通信以及调用厂家提供的api及sdk，编译后的可执行文件名字为syncd）与swss-common，swss以后有机会的话再讲吧。

saireids地址：Azure/sonic-sairedis at 13474d17435d3876e7bd6b50133d25bb11dd3c54 (github.com)

在sonic中，database容器中运行的是redis数据库，而在syncd容器中使用了hiredis（redis数据库的c接口）与redis进行通信，hiredis的github地址如下：

https://github.com/search?q=hiredis

1.m_contextConfig初始化

在sonic-sairedis\syncd\Syncd.cpp的syncd构造函数中调用了如下函数，对容器的上下文进行初始化：

auto ccc = sairedis::ContextConfigContainer::loadFromFile(m_commandLineOptions->m_contextConfig.c_str());m_contextConfig = ccc->get(m_commandLineOptions->m_globalContext);

m_commandLineOptions->m_contextConfig.c_str()是启动syncd时指定的命令行参数，如果不指定，将使用默认配置，代码不展开，直接给出默认情况下一些关键变量的值，若有兴趣可自行查看代码

2.连接redis，切换到指定数据库

m_dbAsic = std::make_shared<swss::DBConnector>(m_contextConfig->m_dbAsic, 0);

调用swss::DBConnector的构造函数，该 class 定义如下：

explicit DBConnector(const DBConnector &other);
DBConnector(int dbId, const RedisContext &ctx);
DBConnector(int dbId, const std::string &hostname, int port, unsigned int timeout);
DBConnector(int dbId, const std::string &unixPath, unsigned int timeout);
DBConnector(const std::string &dbName, unsigned int timeout, bool isTcpConn = false);
DBConnector(const std::string &dbName, unsigned int timeout, bool isTcpConn, const std::string &netns);

可以看到，这是一个多态的构造函数，根据传参可知，调用了第三个构造函数，并且bool isTcpConn = false。代码如下：

#define EMPTY_NAMESPACE std::string()//在别处定义DBConnector::DBConnector(const string& dbName, unsigned int timeout, bool isTcpConn, const string& netns): m_dbId(SonicDBConfig::getDbId(dbName, netns)), m_dbName(dbName), m_namespace(netns)
{struct timeval tv = {0, (suseconds_t)timeout * 1000};struct timeval *ptv = timeout ? &tv : NULL;if (isTcpConn){initContext(SonicDBConfig::getDbHostname(dbName, netns).c_str(), SonicDBConfig::getDbPort(dbName, netns), ptv);}else{initContext(SonicDBConfig::getDbSock(dbName, netns).c_str(), ptv);}select(this);
}DBConnector::DBConnector(const string& dbName, unsigned int timeout, bool isTcpConn): DBConnector(dbName, timeout, isTcpConn, EMPTY_NAMESPACE)
{// Empty constructor
}

timeout为0，所以*ptv为NULL。

由于isTcpConn = false，所以执行else分支下的代码。先来看下initContext函数：

void RedisContext::initContext(const char *path, const timeval *tv)
{if (tv){m_conn = redisConnectUnixWithTimeout(path, *tv);}else{m_conn = redisConnectUnix(path);}if (m_conn->err)throw system_error(make_error_code(errc::address_not_available),"Unable to connect to redis (unix-socket)");
}

redisConnectUnix(path)为hiredis提供的api，用以连接redis，*path为SonicDBConfig::getDbSock(dbName, netns).c_str()，dbname为"ASIC_DB"，netns为空字符串。

看一下getDbSock，代码如下：

string SonicDBConfig::getDbSock(const string &dbName, const string &netns)
{return getRedisInfo(dbName, netns).unixSocketPath;
}RedisInstInfo& SonicDBConfig::getRedisInfo(const std::string &dbName, const std::string &netns)
{std::lock_guard<std::recursive_mutex> guard(m_db_info_mutex);SWSS_LOG_ENTER();if (!m_init)//值为falseinitialize(DEFAULT_SONIC_DB_CONFIG_FILE);//余下部分省略
}

DEFAULT_SONIC_DB_CONFIG_FILE为宏定义，值在SonicDBConfig类中定义：

static constexpr const char *DEFAULT_SONIC_DB_CONFIG_FILE = "/var/run/
redis/sonic-db/database_config.json";

该路径是sonic系统中的路径，编译过程出会将sonic-swss-common\common\database_config.json放入sonic系统下的/var/run/
redis下，文件如下：

{"INSTANCES": {"redis":{"hostname" : "127.0.0.1","port" : 6379,"unix_socket_path" : "/var/run/redis/redis.sock"},"redis_chassis":{"hostname" : "redis_chassis.server","port" : 6380,"unix_socket_path" : "/var/run/redis/redis_chassis.sock"}},"DATABASES" : {"APPL_DB" : {"id" : 0,"separator": ":","instance" : "redis"},"ASIC_DB" : {"id" : 1,"separator": ":","instance" : "redis"},"COUNTERS_DB" : {"id" : 2,"separator": ":","instance" : "redis"},"LOGLEVEL_DB" : {"id" : 3,"separator": ":","instance" : "redis"},"CONFIG_DB" : {"id" : 4,"separator": "|","instance" : "redis"},"PFC_WD_DB" : {"id" : 5,"separator": ":","instance" : "redis"},"FLEX_COUNTER_DB" : {"id" : 5,"separator": ":","instance" : "redis"},"STATE_DB" : {"id" : 6,"separator": "|","instance" : "redis"},"SNMP_OVERLAY_DB" : {"id" : 7,"separator": "|","instance" : "redis"},"RESTAPI_DB" : {"id" : 8,"separator": "|","instance" : "redis"},"GB_ASIC_DB" : {"id" : 9,"separator": "|","instance" : "redis"},"GB_COUNTERS_DB" : {"id" : 10,"separator": "|","instance" : "redis"},"GB_FLEX_COUNTER_DB" : {"id" : 11,"separator": "|","instance" : "redis"},"CHASSIS_APP_DB" : {"id" : 12,"separator": "|","instance" : "redis_chassis"},"CHASSIS_STATE_DB" : {"id" : 13,"separator": "|","instance" : "redis_chassis"}},"VERSION" : "1.0"
}

综上所述，initContext函数连接了/var/run/redis/redis.sock，接下来看select(this)，代码如下

void DBConnector::select(DBConnector *db)
{string select("SELECT ");select += to_string(db->getDbId());RedisReply r(db, select, REDIS_REPLY_STATUS);r.checkStatusOK();
}

顾名思义，改函数执行了redis命令中的SELECT命令，切换到ASIC数据库。RedisReply封装了hiredis中的api，此处不展开。