序

本文主要研究下flink的checkpoint配置

例項

StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();

// start a checkpoint every 1000 ms
env.enableCheckpointing(1000);

// advanced options:

// set mode to exactly-once (this is the default)
env.getCheckpointConfig().setCheckpointingMode(CheckpointingMode.EXACTLY_ONCE);

// checkpoints have to complete within one minute, or are discarded
env.getCheckpointConfig().setCheckpointTimeout(60000);

// make sure 500 ms of progress happen between checkpoints
env.getCheckpointConfig().setMinPauseBetweenCheckpoints(500);

// allow only one checkpoint to be in progress at the same time
env.getCheckpointConfig().setMaxConcurrentCheckpoints(1);

// enable externalized checkpoints which are retained after job cancellation
env.getCheckpointConfig().enableExternalizedCheckpoints(ExternalizedCheckpointCleanup.RETAIN_ON_CANCELLATION);

// This determines if a task will be failed if an error occurs in the execution of the task’s checkpoint procedure.
env.getCheckpointConfig().setFailOnCheckpointingErrors(true);
 

• 使用StreamExecutionEnvironment.enableCheckpointing方法來設定開啟checkpoint；具體可以使用enableCheckpointing(long interval)，或者enableCheckpointing(long interval, CheckpointingMode mode)；interval用於指定checkpoint的觸發間隔(單位milliseconds)，而CheckpointingMode預設是CheckpointingMode.EXACTLY_ONCE，也可以指定為CheckpointingMode.AT_LEAST_ONCE
• 也可以通過StreamExecutionEnvironment.getCheckpointConfig().setCheckpointingMode來設定CheckpointingMode，一般對於超低延遲的應用(大概幾毫秒)可以使用CheckpointingMode.AT_LEAST_ONCE，其他大部分應用使用CheckpointingMode.EXACTLY_ONCE就可以
• checkpointTimeout用於指定checkpoint執行的超時時間(單位milliseconds)，超時沒完成就會被abort掉
• minPauseBetweenCheckpoints用於指定checkpoint coordinator上一個checkpoint完成之後最小等多久可以出發另一個checkpoint，當指定這個引數時，maxConcurrentCheckpoints的值為1
• maxConcurrentCheckpoints用於指定執行中的checkpoint最多可以有多少個，用於包裝topology不會花太多的時間在checkpoints上面；如果有設定了minPauseBetweenCheckpoints，則maxConcurrentCheckpoints這個引數就不起作用了(大於1的值不起作用)
• enableExternalizedCheckpoints用於開啟checkpoints的外部持久化，但是在job失敗的時候不會自動清理，需要自己手工清理state；ExternalizedCheckpointCleanup用於指定當job canceled的時候externalized checkpoint該如何清理，DELETE_ON_CANCELLATION的話，在job canceled的時候會自動刪除externalized state，但是如果是FAILED的狀態則會保留；RETAIN_ON_CANCELLATION則在job canceled的時候會保留externalized checkpoint state
• failOnCheckpointingErrors用於指定在checkpoint發生異常的時候，是否應該fail該task，預設為true，如果設定為false，則task會拒絕checkpoint然後繼續執行

flink-conf.yaml相關配置

#==============================================================================
# Fault tolerance and checkpointing
#==============================================================================

# The backend that will be used to store operator state checkpoints if
# checkpointing is enabled.
#
# Supported backends are 'jobmanager', 'filesystem', 'rocksdb', or the
# <class-name-of-factory>.
#
# state.backend: filesystem

# Directory for checkpoints filesystem, when using any of the default bundled
# state backends.
#
# state.checkpoints.dir: hdfs://namenode-host:port/flink-checkpoints

# Default target directory for savepoints, optional.
#
# state.savepoints.dir: hdfs://namenode-host:port/flink-checkpoints

# Flag to enable/disable incremental checkpoints for backends that
# support incremental checkpoints (like the RocksDB state backend). 
#
# state.backend.incremental: false

• state.backend用於指定checkpoint state儲存的backend，預設為none
• state.backend.async用於指定backend是否使用非同步snapshot(預設為true)，有些不支援async或者只支援async的state backend可能會忽略這個引數
• state.backend.fs.memory-threshold，預設為1024，用於指定儲存於files的state大小閾值，如果小於該值則會儲存在root checkpoint metadata file
• state.backend.incremental，預設為false，用於指定是否採用增量checkpoint，有些不支援增量checkpoint的backend會忽略該配置
• state.backend.local-recovery，預設為false
• state.checkpoints.dir，預設為none，用於指定checkpoint的data files和meta data儲存的目錄，該目錄必須對所有參與的TaskManagers及JobManagers可見
• state.checkpoints.num-retained，預設為1，用於指定保留的已完成的checkpoints個數
• state.savepoints.dir，預設為none，用於指定savepoints的預設目錄
• taskmanager.state.local.root-dirs，預設為none

小結

• 可以通過使用StreamExecutionEnvironment.enableCheckpointing方法來設定開啟checkpoint；具體可以使用enableCheckpointing(long interval)，或者enableCheckpointing(long interval, CheckpointingMode mode)
• checkpoint的高階配置可以配置checkpointTimeout(用於指定checkpoint執行的超時時間，單位milliseconds)，minPauseBetweenCheckpoints(用於指定checkpoint coordinator上一個checkpoint完成之後最小等多久可以出發另一個checkpoint)，maxConcurrentCheckpoints(用於指定執行中的checkpoint最多可以有多少個，如果有設定了minPauseBetweenCheckpoints，則maxConcurrentCheckpoints這個引數大於1的值不起作用)，enableExternalizedCheckpoints(用於開啟checkpoints的外部持久化，在job failed的時候externalized checkpoint state無法自動清理，但是在job canceled的時候可以配置是刪除還是保留state)
• 在flink-conf.yaml裡頭也有checkpoint的相關配置，主要是state backend的配置，比如state.backend.async、state.backend.incremental、state.checkpoints.dir、state.savepoints.dir等

doc

• Checkpointing