Flink MapState的ConcurrentModificationException問題

直接上代碼

import com.alibaba.fastjson.JSON;
import com.tc.flink.analysis.label.bean.output.ItemIdWithAction;
import org.apache.flink.api.common.functions.RichMapFunction;
import org.apache.flink.api.common.state.*;
import org.apache.flink.api.common.time.Time;
import org.apache.flink.api.java.tuple.Tuple2;
import org.apache.flink.configuration.Configuration;

import java.util.HashMap;
import java.util.Map;

public class CityItemIdClickedState extends RichMapFunction<Tuple2<ItemIdWithAction, Integer>, Tuple2<String, String>> {

    private transient MapState<ItemIdWithAction, Integer> map;

    public transient static final String CLICK_PREFIX_KEY = "cityClicked";

    public transient static final String CREATE_PREFIX_KEY = "cityCreated";

    @Override
    public void open(Configuration parameters) throws Exception {
        StateTtlConfig ttlConfig = StateTtlConfig.newBuilder(Time.minutes(60 * 2)).setUpdateType(StateTtlConfig.UpdateType.OnCreateAndWrite).setStateVisibility(StateTtlConfig.StateVisibility.NeverReturnExpired).build();
        MapStateDescriptor<ItemIdWithAction, Integer> descriptor = new MapStateDescriptor<ItemIdWithAction, Integer>("paln_click_num", ItemIdWithAction.class, Integer.class);
        descriptor.enableTimeToLive(ttlConfig);
        map = getRuntimeContext().getMapState(descriptor);
        super.open(parameters);
    }

    @Override
    public Tuple2<String, String> map(Tuple2<ItemIdWithAction, Integer> keyValue) throws Exception {
        Integer num = keyValue.f1;
        ItemIdWithAction itemIdWithAction = keyValue.f0;
        if (num.equals(map.get(itemIdWithAction))) {
            return Tuple2.of(null, null);
        }
        map.put(itemIdWithAction, num);
        String prefixKey = itemIdWithAction.getAction().equals("click") ? CLICK_PREFIX_KEY : CREATE_PREFIX_KEY;
        String key = String.format("%s@%s@%s", prefixKey, itemIdWithAction.getStartCityId(), itemIdWithAction.getEndCityId());
        Map<String, Integer> valueMap = new HashMap<String, Integer>();
        for (ItemIdWithAction tmp : map.keys()) { //報錯的異常點
            valueMap.put(tmp.getItemId(), map.get(tmp));
        }
        return Tuple2.of(key, JSON.toJSONString(valueMap));
    }
}

map是一條條處理,每次取所有MapState數據輸出。
本地跑,集羣跑都沒問題,但當流量大MapState過大時候,就報如下錯誤,每天報錯一兩次,重啓。

java.util.ConcurrentModificationException
	at java.util.HashMap$HashIterator.nextNode(HashMap.java:1442)
	at java.util.HashMap$EntryIterator.next(HashMap.java:1476)
	at java.util.HashMap$EntryIterator.next(HashMap.java:1474)
	at org.apache.flink.runtime.state.ttl.TtlMapState$EntriesIterator.hasNext(TtlMapState.java:161)
	at com.tc.flink.operator.state.CityItemIdClickedState.map(CityItemIdClickedState.java:42)
	at com.tc.flink.operator.state.CityItemIdClickedState.map(CityItemIdClickedState.java:14)
	at org.apache.flink.streaming.api.operators.StreamMap.processElement(StreamMap.java:41)
	at org.apache.flink.streaming.runtime.io.StreamInputProcessor.processInput(StreamInputProcessor.java:202)
	at org.apache.flink.streaming.runtime.tasks.OneInputStreamTask.run(OneInputStreamTask.java:105)
	at org.apache.flink.streaming.runtime.tasks.StreamTask.invoke(StreamTask.java:302)
	at org.apache.flink.runtime.taskmanager.Task.run(Task.java:711)
	at java.lang.Thread.run(Thread.java:748)

報錯原因在於 for (ItemIdWithAction tmp : map.keys()) mapstate被併發修改了。
比較奇怪的是map-function單線程處理,爲什麼出現ConcurrentModificationException
查看TtlMapState源碼
在這裏插入圖片描述
originalIterator::remove是剔除動作。有點類似redis,當再次訪問時候,纔會觸發剔除(有可能產生內存泄漏)。
但是map-funciton是key-by下單線程操作,爲什麼會出現併發問題。
再看TtlStateFactory類
在這裏插入圖片描述

確實異步刪除,所以mapstate過大的時候,就會出現這種問題。
修改代碼

      Iterator< Map.Entry<ItemAction,Integer>> mapIterator= map.iterator();
      while(mapIterator.hasNext()){
            Map.Entry<ItemIdWithAction,Integer>  entry= mapIterator.next();
            valueMap.put(entry.getKey().getItemId(),entry.getValue());
        }

犯了低級錯誤,不過也細讀了源碼

發表評論
所有評論
還沒有人評論,想成為第一個評論的人麼? 請在上方評論欄輸入並且點擊發布.
相關文章