源码阅读之storm操作zookeeper-cluster.clj

storm操作zookeeper的主要函数都定义在命名空间backtype.storm.cluster中（即cluster.clj文件中）。 backtype.storm.cluster定义了两个重要protocol：ClusterState和StormClusterState。

clojure中的protocol可以看成java中的接口，封装了一组方法。ClusterState协议中封装了一组与zookeeper进行交互的基础函数，如获取子节点函数，获取子节点数据函数等，ClusterState协议定义如下：

ClusterState协议

(defprotocol ClusterState
 (set-ephemeral-node [this path data])
 (delete-node [this path])
 (create-sequential [this path data])
 ;; if node does not exist, create persistent with this data
 (set-data [this path data])
 (get-data [this path watch?])
 (get-version [this path watch?])
 (get-data-with-version [this path watch?])
 (get-children [this path watch?])
 (mkdirs [this path])
 (close [this])
 (register [this callback])
 (unregister [this id]))

StormClusterState协议封装了一组storm与zookeeper进行交互的函数，可以将StormClusterState协议中的函数看成ClusterState协议中函数的"组合"。StormClusterState协议定义如下：

StormClusterState协议

(defprotocol StormClusterState
(assignments [this callback])
(assignment-info [this storm-id callback])
(assignment-info-with-version [this storm-id callback])
(assignment-version [this storm-id callback])
(active-storms [this])
(storm-base [this storm-id callback])
(get-worker-heartbeat [this storm-id node port])
(executor-beats [this storm-id executor->node+port])
(supervisors [this callback])
(supervisor-info [this supervisor-id]) ;; returns nil if doesn't exist
(setup-heartbeats! [this storm-id])
(teardown-heartbeats! [this storm-id])
(teardown-topology-errors! [this storm-id])
(heartbeat-storms [this])
(error-topologies [this])
(worker-heartbeat! [this storm-id node port info])
(remove-worker-heartbeat! [this storm-id node port])
(supervisor-heartbeat! [this supervisor-id info])
(activate-storm! [this storm-id storm-base])
(update-storm! [this storm-id new-elems])
(remove-storm-base! [this storm-id])
(set-assignment! [this storm-id info])
(remove-storm! [this storm-id])
(report-error [this storm-id task-id node port error])
(errors [this storm-id task-id])
(disconnect [this]))

命名空间backtype.storm.cluster除了定义ClusterState和StormClusterState这两个重要协议外，还定义了两个重要函数：mk-distributed-cluster-state和mk-storm-cluster-state。

mk-distributed-cluster-state函数如下：

该函数返回一个实现了ClusterState协议的对象，通过这个对象就可以与zookeeper进行交互了。

mk-distributed-cluster-state函数

(defn mk-distributed-cluster-state
;; conf绑定了storm.yaml中的配置信息，是一个map对象
[conf]
;; zk绑定一个zk client，Storm使用CuratorFramework与Zookeeper进行交互
(let [zk (zk/mk-client conf (conf STORM-ZOOKEEPER-SERVERS) (conf STORM-ZOOKEEPER-PORT)            :auth-conf conf)]
 ;; 创建storm集群在zookeeper上的根目录，默认值为/storm
 (zk/mkdirs zk (conf STORM-ZOOKEEPER-ROOT))
 (.close zk))
;; callbacks绑定回调函数集合，是一个map对象
(let [callbacks (atom {})
 ;; active标示zookeeper集群状态
 active (atom true)
 ;; zk重新绑定新的zk client，该zk client设置了watcher，这样当zookeeper集群的状态发生变化时，zk server会给zk client发送相应的event，zk client设置的watcher会调用callbacks中相应回调函数来处理event
 ;; 启动nimbus时，callbacks是一个空集合，所以nimbus端收到event后不会调用任何回调函数；但是启动supervisor时，callbacks中注册了回调函数，所以当supervisor收到zk server发送的event后，会调用相应的回调函数
 ;; mk-client函数定义在zookeeper.clj文件中，请参见其定义部分
 zk (zk/mk-client conf
          (conf STORM-ZOOKEEPER-SERVERS)
          (conf STORM-ZOOKEEPER-PORT)
          :auth-conf conf
          :root (conf STORM-ZOOKEEPER-ROOT)
          ;; :watcher绑定一个函数，指定zk client的默认watcher函数，state标示当前zk client的状态；type标示事件类型；path标示zookeeper上产生该事件的znode
          ;; 该watcher函数主要功能就是执行callbacks集合中的函数，callbacks集合中的函数是在mk-storm-cluster-state函数中通过调用ClusterState的register函数添加的
          :watcher (fn [state type path]
               (when @active
                (when-not (= :connected state)
                 (log-warn "Received event " state ":" type ":" path " with disconnected Zookeeper."))
                (when-not (= :none type)
                 (doseq [callback (vals @callbacks)]
                  (callback type path))))))]
;; reify相当于java中的implements，这里表示实现一个协议
(reify
ClusterState
;; register函数用于将回调函数加入callbacks中，key是一个32位的标识
(register
 [this callback]
 (let [id (uuid)]
  (swap! callbacks assoc id callback)
  id))
;; unregister函数用于将指定key的回调函数从callbacks中删除
(unregister
 [this id]
 (swap! callbacks dissoc id))
;; 在zookeeper上添加一个临时节点
(set-ephemeral-node
 [this path data]
 (zk/mkdirs zk (parent-path path))
 (if (zk/exists zk path false)
  (try-cause
   (zk/set-data zk path data) ; should verify that it's ephemeral
   (catch KeeperException$NoNodeException e
    (log-warn-error e "Ephemeral node disappeared between checking for existing and setting data")
    (zk/create-node zk path data :ephemeral)
    ))
  (zk/create-node zk path data :ephemeral)))
;; 在zookeeper上添加一个顺序节点
(create-sequential
 [this path data]
 (zk/create-node zk path data :sequential))
;; 修改某个节点数据
(set-data
 [this path data]
 ;; note: this does not turn off any existing watches
 (if (zk/exists zk path false)
  (zk/set-data zk path data)
  (do
   (zk/mkdirs zk (parent-path path))
   (zk/create-node zk path data :persistent))))
;; 删除指定节点
(delete-node
 [this path]
 (zk/delete-recursive zk path))
;; 获取指定节点数据。path标示节点路径；watch?是一个布尔类型值，表示是否需要对该节点进行"观察"，如果watch?=true，当调用set-data函数修改该节点数据后，
;; 会给zk client发送一个事件，zk client接收事件后，会调用创建zk client时指定的默认watcher函数（即:watcher绑定的函数）
(get-data
 [this path watch?]
 (zk/get-data zk path watch?))
;; 与get-data函数的区别就是获取指定节点数据的同时，获取节点数据的version，version表示节点数据修改的次数
(get-data-with-version
 [this path watch?]
 (zk/get-data-with-version zk path watch?))
;; 获取指定节点的version，watch?的含义与get-data函数中的watch?相同
(get-version
 [this path watch?]
 (zk/get-version zk path watch?))
;; 获取指定节点的子节点列表，watch?的含义与get-data函数中的watch?相同
(get-children
 [this path watch?]
 (zk/get-children zk path watch?))
;; 在zookeeper上创建一个节点
(mkdirs
 [this path]
 (zk/mkdirs zk path))
;; 关闭zk client
(close
 [this]
 (reset! active false)
 (.close zk)))))

mk-storm-cluster-state函数定义如下：

mk-storm-cluster-state函数非常重要，该函数返回一个实现了StormClusterState协议的实例，通过该实例storm就可以更加方便与zookeeper进行交互。

在启动nimbus和supervisor的函数中均调用了mk-storm-cluster-state函数。关于nimbus和supervisor的启动将在之后的文章中介绍。

mk-storm-cluster-state函数

(defn mk-storm-cluster-state
[cluster-state-spec]
;; satisfies?谓词相当于java中的instanceof，判断cluster-state-spec是不是ClusterState实例
(let [[solo? cluster-state] (if (satisfies? ClusterState cluster-state-spec)
             [false cluster-state-spec]
             [true (mk-distributed-cluster-state cluster-state-spec)])
 ;; 绑定topology id->回调函数的map，当/assignments/{topology id}数据发生变化时，zk client执行assignment-info-callback中topology id所对应的回调函数
 assignment-info-callback (atom {})
 ;; assignment-info-with-version-callback与assignment-info-callback类似
 assignment-info-with-version-callback (atom {})
 ;; assignment-version-callback与assignments-callback类似
 assignment-version-callback (atom {})
 ;; 当/supervisors标示的znode的子节点发生变化时，zk client执行supervisors-callback指向的函数
 supervisors-callback (atom nil)
 ;; 当/assignments标示的znode的子节点发生变化时，zk client执行assignments-callback指向的函数
 assignments-callback (atom nil)
 ;; 当/storms/{topology id}标示的znode的数据发生变化时，zk client执行storm-base-callback中topology id所对应的回调函数
 storm-base-callback (atom {})
 ;; register函数将"回调函数(fn ...)"添加到cluster-state的callbacks集合中，并返回标示该回调函数的uuid
 state-id (register
       cluster-state
       ;; 定义"回调函数"，type标示事件类型，path标示znode
       (fn [type path]
        ;; subtree绑定路径前缀如"assignments"、"storms"、"supervisors"等，args存放topology id
        (let [[subtree & args] (tokenize-path path)]
         ;; condp相当于java中的switch
         (condp = subtree
          ;; 当subtree="assignments"时，如果args为空，说明是/assignments的子节点发生变化，执行assignments-callback指向的回调函数，否则
    ;; 说明/assignments/{topology id}标示的节点数据发生变化，执行assignment-info-callback指向的回调函数
          ASSIGNMENTS-ROOT (if (empty? args)
                   (issue-callback! assignments-callback)
                   (issue-map-callback! assignment-info-callback (first args)))
          ;; 当subtree="supervisors"时，说明是/supervisors的子节点发生变化，执行supervisors-callback指向的回调函数
          SUPERVISORS-ROOT (issue-callback! supervisors-callback)
          ;; 当subtree="storms"时，说明是/storms/{topology id}标示的节点数据发生变化，执行storm-base-callback指向的回调函数
          STORMS-ROOT (issue-map-callback! storm-base-callback (first args))
          ;; this should never happen
          (exit-process! 30 "Unknown callback for subtree " subtree args)))))]
;; 在zookeeper上创建storm运行topology所必需的znode
(doseq [p [ASSIGNMENTS-SUBTREE STORMS-SUBTREE SUPERVISORS-SUBTREE WORKERBEATS-SUBTREE ERRORS-SUBTREE]]
(mkdirs cluster-state p))
;; 返回一个实现StormClusterState协议的实例
(reify
StormClusterState
;; 获取/assignments的子节点列表，如果callback不为空，将其赋值给assignments-callback，并对/assignments添加"节点观察"
(assignments
 [this callback]
 (when callback
  (reset! assignments-callback callback))
 (get-children cluster-state ASSIGNMENTS-SUBTREE (not-nil? callback)))
;; 获取/assignments/{storm-id}节点数据，即storm-id的分配信息，如果callback不为空，将其添加到assignment-info-callback中，并对/assignments/{storm-id}添加"数据观察"
(assignment-info
 [this storm-id callback]
 (when callback
  (swap! assignment-info-callback assoc storm-id callback))
 (maybe-deserialize (get-data cluster-state (assignment-path storm-id) (not-nil? callback))))
;; 获取/assignments/{storm-id}节点数据包括version信息，如果callback不为空，将其添加到assignment-info-with-version-callback中，并对/assignments/{storm-id}添加"数据观察"
(assignment-info-with-version
 [this storm-id callback]
 (when callback
  (swap! assignment-info-with-version-callback assoc storm-id callback))
 (let [{data :data version :version}
    (get-data-with-version cluster-state (assignment-path storm-id) (not-nil? callback))]
 {:data (maybe-deserialize data)
  :version version}))
;; 获取/assignments/{storm-id}节点数据的version信息，如果callback不为空，将其添加到assignment-version-callback中，并对/assignments/{storm-id}添加"数据观察"
(assignment-version
 [this storm-id callback]
 (when callback
  (swap! assignment-version-callback assoc storm-id callback))
 (get-version cluster-state (assignment-path storm-id) (not-nil? callback)))
;; 获取storm集群中正在运行的topology id即/storms的子节点列表
(active-storms
 [this]
 (get-children cluster-state STORMS-SUBTREE false))
;; 获取storm集群中所有有心跳的topology id即/workerbeats的子节点列表
(heartbeat-storms
 [this]
 (get-children cluster-state WORKERBEATS-SUBTREE false))
;; 获取所有有错误的topology id即/errors的子节点列表
(error-topologies
 [this]
 (get-children cluster-state ERRORS-SUBTREE false))
;; 获取指定storm-id进程的心跳信息，即/workerbeats/{storm-id}/{node-port}节点数据
(get-worker-heartbeat
 [this storm-id node port]
 (-> cluster-state
   (get-data (workerbeat-path storm-id node port) false)
   maybe-deserialize))
;; 获取指定进程中所有线程的心跳信息
(executor-beats
 [this storm-id executor->node+port]
 ;; need to take executor->node+port in explicitly so that we don't run into a situation where a
 ;; long dead worker with a skewed clock overrides all the timestamps. By only checking heartbeats
 ;; with an assigned node+port, and only reading executors from that heartbeat that are actually assigned,
 ;; we avoid situations like that
 (let [node+port->executors (reverse-map executor->node+port)
    all-heartbeats (for [[[node port] executors] node+port->executors]
             (->> (get-worker-heartbeat this storm-id node port)
               (convert-executor-beats executors)
               ))]
  (apply merge all-heartbeats)))
;; 获取/supervisors的子节点列表，如果callback不为空，将其赋值给supervisors-callback，并对/supervisors添加"节点观察"
(supervisors
 [this callback]
 (when callback
  (reset! supervisors-callback callback))
 (get-children cluster-state SUPERVISORS-SUBTREE (not-nil? callback)))
;; 获取/supervisors/{supervisor-id}节点数据，即supervisor的心跳信息
(supervisor-info
 [this supervisor-id]
 (maybe-deserialize (get-data cluster-state (supervisor-path supervisor-id) false)))
;; 设置进程心跳信息
(worker-heartbeat!
 [this storm-id node port info]
 (set-data cluster-state (workerbeat-path storm-id node port) (Utils/serialize info)))
;; 删除进程心跳信息
(remove-worker-heartbeat!
 [this storm-id node port]
 (delete-node cluster-state (workerbeat-path storm-id node port)))
;; 创建指定storm-id的topology的用于存放心跳信息的节点
(setup-heartbeats!
 [this storm-id]
 (mkdirs cluster-state (workerbeat-storm-root storm-id)))
;; 删除指定storm-id的topology的心跳信息节点
(teardown-heartbeats!
 [this storm-id]
 (try-cause
  (delete-node cluster-state (workerbeat-storm-root storm-id))
  (catch KeeperException e
   (log-warn-error e "Could not teardown heartbeats for " storm-id))))
;; 删除指定storm-id的topology的错误信息节点
(teardown-topology-errors!
 [this storm-id]
 (try-cause
  (delete-node cluster-state (error-storm-root storm-id))
  (catch KeeperException e
   (log-warn-error e "Could not teardown errors for " storm-id))))
;; 创建临时节点存放supervisor的心跳信息
(supervisor-heartbeat!
 [this supervisor-id info]
 (set-ephemeral-node cluster-state (supervisor-path supervisor-id) (Utils/serialize info)))
;; 创建/storms/{storm-id}节点
(activate-storm!
 [this storm-id storm-base]
 (set-data cluster-state (storm-path storm-id) (Utils/serialize storm-base)))
;; 更新topology对应的StormBase对象，即更新/storm/{storm-id}节点
(update-storm!
 [this storm-id new-elems]
 ;; base绑定storm-id在zookeeper上的StormBase对象
 (let [base (storm-base this storm-id nil)
    ;; executors绑定component名称->组件并行度的map
    executors (:component->executors base)
    ;; new-elems绑定合并后的组件并行度map，update函数将组件新并行度map合并到旧map中
    new-elems (update new-elems :component->executors (partial merge executors))]
  ;; 更新StormBase对象中的组件并行度map，并写入zookeeper的/storms/{storm-id}节点
  (set-data cluster-state (storm-path storm-id)
       (-> base
         (merge new-elems)
         Utils/serialize))))
;; 获取storm-id的StormBase对象，即读取/storms/{storm-id}节点数据，如果callback不为空，将其赋值给storm-base-callback，并为/storms/{storm-id}节点添加"数据观察"
(storm-base
 [this storm-id callback]
 (when callback
  (swap! storm-base-callback assoc storm-id callback))
 (maybe-deserialize (get-data cluster-state (storm-path storm-id) (not-nil? callback))))
;; 删除storm-id的StormBase对象，即删除/storms/{storm-id}节点
(remove-storm-base!
 [this storm-id]
 (delete-node cluster-state (storm-path storm-id)))
;; 更新storm-id的分配信息，即更新/assignments/{storm-id}节点数据
(set-assignment!
 [this storm-id info]
 (set-data cluster-state (assignment-path storm-id) (Utils/serialize info)))
;; 删除storm-id的分配信息，同时删除其StormBase信息，即删除/assignments/{storm-id}节点和/storms/{storm-id}节点
(remove-storm!
 [this storm-id]
 (delete-node cluster-state (assignment-path storm-id))
 (remove-storm-base! this storm-id))
;; 将组件异常信息写入zookeeper
(report-error
 [this storm-id component-id node port error]
 ;; path绑定"/errors/{storm-id}/{component-id}"
 (let [path (error-path storm-id component-id)
    ;; data绑定异常信息，包括异常时间、异常堆栈信息、主机和端口
    data {:time-secs (current-time-secs) :error (stringify-error error) :host node :port port}
    ;; 创建/errors/{storm-id}/{component-id}节点
    _ (mkdirs cluster-state path)
    ;; 创建/errors/{storm-id}/{component-id}的子顺序节点，并写入异常信息
    _ (create-sequential cluster-state (str path "/e") (Utils/serialize data))
    ;; to-kill绑定除去顺序节点编号最大的前10个节点的剩余节点的集合
    to-kill (->> (get-children cluster-state path false)
           (sort-by parse-error-path)
           reverse
           (drop 10))]
  ;; 删除to-kill中包含的节点
  (doseq [k to-kill]
   (delete-node cluster-state (str path "/" k)))))
;; 得到给定的storm-id component-id下的异常信息
(errors
 [this storm-id component-id]
 (let [path (error-path storm-id component-id)
    _ (mkdirs cluster-state path)
    children (get-children cluster-state path false)
    errors (dofor [c children]
           (let [data (-> (get-data cluster-state (str path "/" c) false)
                   maybe-deserialize)]
            (when data
             (struct TaskError (:error data) (:time-secs data) (:host data) (:port data))
             )))
    ]
  (->> (filter not-nil? errors)
     (sort-by (comp - :time-secs)))))
;; 关闭连接，在关闭连接前，将回调函数从cluster-state的callbacks中删除
(disconnect
 [this]
 (unregister cluster-state state-id)
 (when solo?
  (close cluster-state))))))

zookeeper.clj中mk-client函数

mk-client函数创建一个CuratorFramework实例，为该实例注册了CuratorListener，当一个后台操作完成或者指定的watch被触发时将会执行CuratorListener中的eventReceived()。eventReceived中调用的wacher函数就是mk-distributed-cluster-state中:watcher绑定的函数。

(defnk mk-client
[conf servers port
 :root ""
 :watcher default-watcher
 :auth-conf nil]
(let [fk (Utils/newCurator conf servers port root (when auth-conf (ZookeeperAuthInfo. auth-conf)))]
 (.. fk
   (getCuratorListenable)
   (addListener
    (reify CuratorListener
     (^void eventReceived [this ^CuratorFramework _fk ^CuratorEvent e]
       (when (= (.getType e) CuratorEventType/WATCHED)
        (let [^WatchedEvent event (.getWatchedEvent e)]
         (watcher (zk-keeper-states (.getState event))
             (zk-event-types (.getType event))
             (.getPath event))))))))
 (.start fk)
 fk))

mk-distributed-cluster-state函数创建了一个zk client，并通过:watcher给该zk client指定了"wacher"函数，这个"wacher"函数只是简单调用ClusterState的callbacks集合中的函数，这样这个"wacher"函数执行 哪些函数将由ClusterState实例决定
ClusterState实例提供register函数来更新callbacks集合，ClusterState实例被传递给了mk-storm-cluster-state函数，在mk-storm-cluster-state中调用register添加了一个函数(fn [type path] ... )，这个函数实现了"watcher"函数的全部逻辑
mk-storm-cluster-state中注册的函数执行的具体内容由StormClusterState实例决定，对zookeeper节点添加"观察"也是通过StormClusterState实例实现的，这样我们就可以通过StormClusterState实例对我们感兴趣的节点添加"观察"和"回调函数"，当节点或节点数据发生变化后，zk server就会给zk client发送"通知"，zk client中的"wather"函数将被调用，进而我们注册的"回到函数"将被执行。

总结

这部分源码与zookeeper联系十分紧密，涉及了很多zookeeper中的概念和特性，如"数据观察"和"节点观察"等，有关zookeeper的wacher机制请参考：https://www.jb51.net/article/124295.htm，storm并没有直接使用zookeeper的api，而是使用Curator框架，Curator框架简化了访问zookeeper的操作。关于Curator框架请参考：https://www.jb51.net/article/125785.htm。

来源：http://blog.csdn.net/mangocream/article/details/42456901