您的位置:首页 > 其它

【zookeeper】 同步API示例入门

2016-02-01 16:48 239 查看
zookeeper提供了很方便的管理工具,同步API也很方便使用,本文主要介绍以下几个API的用法。

1. zookeeper初始化

zhandle_t *zookeeper_init(const char *host, watcher_fn fn, int recv_timeout, const clientid_t *clientid, void *context, int flags);
功能:

创建一个句柄(handle)和一个响应(response)这个句柄的会话(session)。

参数:

host:zookeeper主机列表,用逗号间隔。

fn:用于监视的回调函数。

clientid:之前建立过连接,现在要重新连的客户端(client)ID。如果之前没有,则为0.

context:暂时用不到,忽略。(TODO)

flags:设置为0,zookeeper开发团队保留以后使用。

2. 监视节点是否存在

int zoo_exists(zhandle_t *zh, const char *path, int watch,
struct Stat *stat);

功能:

同步监视一个zookeeper节点(node)是否存在。

参数:

zh:zookeeper的句柄,由zookeeper_init得到。

path:节点名称,就是一个类似于文件系统写法的路径。

watch:设置为0,则无作用。设置为非0时,暂时用不到,忽略。(TODO)

stat:(TODO)

返回值:ZOK,ZNONODE,ZNOAUTH,ZBADARGUMENTS,ZINVALIDSTATE,ZMARSHALLINGERROR。ZOK表示操作成功,ZNONODE表示该节点不存在,ZNOAUTH表示客户端(client)无权限,ZINVALIDSTATE表示存在非法的参数,后两者暂略(TODO)。

3. 新建zookeeper节点

int zoo_create(zhandle_t *zh, const char *path, const char *value, int valuelen, const struct ACL_vector *acl, int flags, char *path_buffer, int path_buffer_len);

功能:

创建一个同步的zookeeper节点。

参数:

zh:zookeeper的句柄,由zookeeper_init得到。

path:节点名称,就是一个类似于文件系统写法的路径。

value:欲存储到该节点的数据。如果不存储数据,则设置为NULL。

valuelen:欲存储的数据的长度。如果不存储数据,则设置为-1.

acl:初始的ACL节点,ACL不能为空。比如设置为&ZOO_OPEN_ACL_UNSAFE。(TODO)

flags:一般设置为0.(TODO)

path_buffer:将由新节点填充的路径值。可设置为NULL。(TODO)

path_buffer_len:path_buffer的长度。

返回值:ZOK,ZNONODE,ZNODEEXISTS,ZNOAUTH,ZNOCHILDRENFOREPHEMERALS,ZBADARGUMENTS,ZINVALIDSTATE,ZMARSHALLINGERROR。ZOK表示操作成功,ZNONODE表示该节点不存在,ZNODEEXISTS表示节点已经存在,ZNOAUTH表示客户端(client)无权限,ZNOCHILDRENFOREPHEMERALS表示不能够创建临时(ephemeral)节点的子节点(children),ZINVALIDSTATE表示存在非法的参数,后两者暂略(TODO)。

4. 设置zookeeper节点

int zoo_set(zhandle_t *zh, const char *path, const char *buffer, int buflen, int version);

功能:

向zookeeper节点写数据。

参数:

zh:zookeeper的句柄,由zookeeper_init得到。

path:节点名称,就是一个类似于文件系统写法的路径。

buffer:欲写的数据。
buflen:欲写的数据的长度。

version:检查这个节点的版本是否为version。当设置为-1时,不会进行版本检查。(TODO)

5. 获取某节点的子节点

int zoo_wget_children(zhandle_t *zh, const char *path, watcher_fn watcher, void* watcherCtx, struct String_vector *strings);

功能:

同步列出一个节点的所有子节点。

参数:

zh:zookeeper的句柄,由zookeeper_init得到。

path:节点名称,就是一个类似于文件系统写法的路径。

watcher:若设置为非0,则zookeeper服务器(server)会设置一个监视器,用来在节点发生改变时通知客户端(client)。

watcherCtx:传送给watcher作为回调的具体用户数据(user specific data)。

strings:用来保存得到的子节点路径。

返回值:ZOK,ZNONODE,ZNOAUTH,ZBADARGUMENTS,ZINVALIDSTATE,ZMARSHALLINGERROR。

在链接的时候需要注意下,官方一共提供了两组库,xxx_st.a和xxx_mt.a,其中xxx_mt.a 支持同步api,链接的时间一定要注意,demo代码如下:

#include <zookeeper.h>
#include <zookeeper_log.h>
#include <iostream>
#include <string.h>

using namespace std;

const char* host = "127.0.0.1:2181";
const int timeout = 2000;
zhandle_t* zkhandle = NULL;
bool is_connect = false;

void clients_watcher_g(zhandle_t * zh, int type, int state, const char* path, void* watcherCtx) {
    printf("global watcher - type:%d,state:%d\n", type, state);
    if ( type==ZOO_SESSION_EVENT ) {
        if ( state==ZOO_CONNECTED_STATE ) {
            printf("connected to zookeeper service successfully!\n");
            printf("timeout:%d\n", zoo_recv_timeout(zh));
            is_connect = true;
        }
        else if ( state==ZOO_EXPIRED_SESSION_STATE ) {
            printf("zookeeper session expired!\n");
        }
    }
    else {
        printf("other type:%d\n", type);
    }
}

int main(int argc, char* argv[])
{
    //Init and connect
    zoo_set_debug_level(ZOO_LOG_LEVEL_WARN);
    if (zkhandle)
        zookeeper_close(zkhandle);

    zkhandle = zookeeper_init(host, clients_watcher_g, timeout, 0, NULL, 0);
    if (NULL == zkhandle) {
        printf("zookeeper init error\n");
        return 0;
    }

    Stat stat;
    int ret = 0;
       char path[] = "/app_watch";
    //check exist
       ret = zoo_exists(zkhandle, path, 0, &stat);
    printf("[%s] ret=%d msg=%s\n", path, ret, ret==0? "succ" : "error");
       if (ZNONODE == ret) {
        // create
        int ret = zoo_create(zkhandle, path, NULL, -1, &ZOO_OPEN_ACL_UNSAFE, 0, NULL, 0);
        printf("create [%s] ret=%d\n", path, ret);
    }

    //check exist
    ret = zoo_exists(zkhandle, "/app_invalid", 0, &stat);
    printf("ret=%d msg=%s\n", ret, ret==0? "succ" : "error");
    char buffer[1024];
    int buff_len;

    //get data
    buff_len = sizeof(buffer);
    bzero(buffer, buff_len);
    ret = zoo_get(zkhandle, "/app1/p_1", 0, buffer, &buff_len, &stat);
    printf("ret=%d stat=%d len=%d buffer=%s\n", ret, stat.ctime, buff_len, buffer);

        //set data
    buff_len = sizeof(buffer);
        snprintf(buffer, buff_len, "\"rank\":\"10.101.1.101\"");
    ret = zoo_set(zkhandle, "/app1/p_1", (const char *)buffer, strlen(buffer), stat.version);
    printf("ret=%d\n", ret);

    zookeeper_close(zkhandle);
}


执行后输出如下:

[/app_watch] ret=0 msg=succ

global watcher - type:-1,state:3

ret=-101 msg=error

connected to zookeeper service successfully!

timeout:4000

ret=0 stat=1591592808 len=21 buffer="rank":"10.101.1.101"

ret=0

附上makefile,方便大家检查链接问题:

BIN = $(patsubst %.cc,%,$(wildcard *.cc))

PATH_ZK = ../zookeeper-3.4.6/src/c/
INC = -I $(PATH_ZK)/include/ -I $(PATH_ZK)/generated
LIB = $(PATH_ZK)/.libs/libzkmt.a $(PATH_ZK)/.libs/libzookeeper_mt.a $(PATH_ZK)/.libs/libhashtable.a
#LIB = $(PATH_ZK)/.libs/libzkst.a $(PATH_ZK)/.libs/libzookeeper_mt.a $(PATH_ZK)/.libs/libhashtable.a

RED = \\e[1m\\e[31m
RESET = \\e[m
GREEN = \\e[1m\\e[32m

all:$(BIN)

%:%.cc
@echo -e "Make $(GREEN)$@$(RESET) begin......\c"
g++ -g -pthread -o $@ $< $(INC) $(LIB)
@echo -e $(RED)"ok."$(RESET)

clean:
rm $(BIN)
@echo "make clean done."


最后附上错误码表和枚举定义,方便自己和大家查阅:

以下定义在zookeeper.h中

enum ZOO_ERRORS {
ZOK = 0, /*!< Everything is OK */

/** System and server-side errors.
* This is never thrown by the server, it shouldn't be used other than
* to indicate a range. Specifically error codes greater than this
* value, but lesser than {@link #ZAPIERROR}, are system errors. */
ZSYSTEMERROR = -1,
ZRUNTIMEINCONSISTENCY = -2, /*!< A runtime inconsistency was found */
ZDATAINCONSISTENCY = -3, /*!< A data inconsistency was found */
ZCONNECTIONLOSS = -4, /*!< Connection to the server has been lost */
ZMARSHALLINGERROR = -5, /*!< Error while marshalling or unmarshalling data */
ZUNIMPLEMENTED = -6, /*!< Operation is unimplemented */
ZOPERATIONTIMEOUT = -7, /*!< Operation timeout */
ZBADARGUMENTS = -8, /*!< Invalid arguments */
ZINVALIDSTATE = -9, /*!< Invliad zhandle state */

/** API errors.
* This is never thrown by the server, it shouldn't be used other than
* to indicate a range. Specifically error codes greater than this
* value are API errors (while values less than this indicate a
* {@link #ZSYSTEMERROR}).
*/
ZAPIERROR = -100,
ZNONODE = -101, /*!< Node does not exist */
ZNOAUTH = -102, /*!< Not authenticated */
ZBADVERSION = -103, /*!< Version conflict */
ZNOCHILDRENFOREPHEMERALS = -108, /*!< Ephemeral nodes may not have children */
ZNODEEXISTS = -110, /*!< The node already exists */
ZNOTEMPTY = -111, /*!< The node has children */
ZSESSIONEXPIRED = -112, /*!< The session has been expired by the server */
ZINVALIDCALLBACK = -113, /*!< Invalid callback specified */
ZINVALIDACL = -114, /*!< Invalid ACL specified */
ZAUTHFAILED = -115, /*!< Client authentication failed */
ZCLOSING = -116, /*!< ZooKeeper is closing */
ZNOTHING = -117, /*!< (not error) no server responses to process */
ZSESSIONMOVED = -118 /*!<session moved to another server, so operation is ignored */
};


以下定义在zookeeper.c中

const int ZOOKEEPER_WRITE = 1 << 0;
const int ZOOKEEPER_READ = 1 << 1;

const int ZOO_EPHEMERAL = 1 << 0;
const int ZOO_SEQUENCE = 1 << 1;

const int ZOO_EXPIRED_SESSION_STATE = EXPIRED_SESSION_STATE_DEF;
const int ZOO_AUTH_FAILED_STATE = AUTH_FAILED_STATE_DEF;
const int ZOO_CONNECTING_STATE = CONNECTING_STATE_DEF;
const int ZOO_ASSOCIATING_STATE = ASSOCIATING_STATE_DEF;
const int ZOO_CONNECTED_STATE = CONNECTED_STATE_DEF;
内容来自用户分享和网络整理,不保证内容的准确性,如有侵权内容,可联系管理员处理 点击这里给我发消息
标签: