redis程序员

Redis scan命令的一次坑

2017-02-04  本文已影响12179人  locoder

Redis作为当前服务架构不可或缺的Cache,其支持丰富多样的数据结构,Redis在使用中其实也有很多坑,本次博主遇到的坑或许说是Java程序员会遇到的多一点,下面就听博主详细道来。

线上服务堵塞

String key = keyOf(appid);
int retryCount = 3;
int socketRetryCount = 3;
Exception ex = null;
while(retryCount > 0 && socketRetryCount > 0) {
    try {
        return redisDao.getMap(key);
    }catch (Exception e) {

    }
}

12月2日被告知服务出现异常,查看日志发现其运行到上述代码getMap方法处后日志就没有内容了。

问题分析

"pool-13-thread-6" prio=10 tid=0x00007f754800e800 nid=0x71b5 waiting on condition [0x00007f758f0ee000]
    java.lang.Thread.State: WAITING (parking)
    at sun.misc.Unsafe.park(Native Method)
    - parking to wait for  <0x0000000779b75f40> (a java.util.concurrent.locks.AbstractQueuedSynchronizer$ConditionObject)
    at java.util.concurrent.locks.LockSupport.park(LockSupport.java:186)
    at java.util.concurrent.locks.AbstractQueuedSynchronizer$ConditionObject.await(AbstractQueuedSynchronizer.java:2043)
    at org.apache.commons.pool2.impl.LinkedBlockingDeque.takeFirst(LinkedBlockingDeque.java:583)
    at org.apache.commons.pool2.impl.GenericObjectPool.borrowObject(GenericObjectPool.java:442)
    at org.apache.commons.pool2.impl.GenericObjectPool.borrowObject(GenericObjectPool.java:363)
    at redis.clients.util.Pool.getResource(Pool.java:49)
    at redis.clients.jedis.JedisPool.getResource(JedisPool.java:99)
    at org.reborndb.reborn.RoundRobinJedisPool.getResource(RoundRobinJedisPool.java:300)
    at com.le.smartconnect.adapter.spring.RebornConnectionFactory.getConnection(RebornConnectionFactory.java:43)
    at org.springframework.data.redis.core.RedisConnectionUtils.doGetConnection(RedisConnectionUtils.java:128)
    at org.springframework.data.redis.core.RedisConnectionUtils.getConnection(RedisConnectionUtils.java:91)
    at org.springframework.data.redis.core.RedisConnectionUtils.getConnection(RedisConnectionUtils.java:78)
    at xxx.run(xxx.java:80)
    at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:471)
    at java.util.concurrent.FutureTask.run(FutureTask.java:262)
    at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
    at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)
at java.lang.Thread.run(Thread.java:745)

    Locked ownable synchronizers:
- <0x000000074f529b08> (a java.util.concurrent.ThreadPoolExecutor$Worker)

从线程日志可以看出服务堵塞在获取redis连接处.

分析:

一次修改分析

从以上两点分析得出,redis连接被耗尽,于是查找代码得知由于重写spring-data-redis中的hscan方面导致,代码如下:

RedisConnection rc = redisTemplate.getConnectionFactory().getConnection();
if (rc instanceof JedisConnection) {
    JedisConnection JedisConnection = (JedisConnection) rc;
    return new ConvertingCursor<Map.Entry<byte[], byte[]>, Map.Entry<String, String>>(
            JedisConnection.hScan(rawValue(key), cursor, scanOptions),
            new Converter<Map.Entry<byte[], byte[]>, Map.Entry<String, String>>() {

            @Override
            public Entry<String, String> convert(final Entry<byte[], byte[]> source) {

                return new Map.Entry<String, String>() {

                @Override
                public String getKey() {
                    return hashKeySerializer.deserialize(source.getKey());
                }

                @Override
                public String getValue() {
                    return hashValueSerializer.deserialize(source.getValue());
                }

                @Override
                public String setValue(String value) {
                    throw new UnsupportedOperationException(
                        "Values cannot be set when scanning through entries.");
                }
            };

        }
    });
} else {
    return hashOps.scan(key, scanOptions);
}

上述代码返回ConvertingCursor后未释放连接,导出连接被占满。

二次修改分析

于是修改代码为正常释放连接

try {
    ...
}finally {
    RedisConnectionUtils.releaseConnection(rc, factory);
}

代码经过上线,再次跑程序查看线上日志发现报了大量的Connection time out.

于是博主就思考是不是由于重写代码不对,尝试使用spring-data-redis的原生代码,即直接调用hashOps.scan(key, scanOptions)方法,再次上线。

上线后观察日志:发现这次不是报Connection time out,日志中大量报Unknown reply:错误。

分析如下:

由于代码是在多线程环境下运行,有几百个线程去调用hscan操作,spring-data-redis原生的代码执行完一次hscan操作后就会关闭连接并返回一个迭代器Cursor,但是遍历Cursor时在本次count后会再次根据游标重新使用该连接进行查询,可是连接却已经被关闭,这时会使用新的连接是可以正常迭代的,但是一旦复用到其他线程使用的连接则会导致报错Unknown reply.

三次修改分析

经过思考后得出结论,redis在执行scan操作时一旦连接被释放,那么scan操作将不会进行下去,则报Connection time out.

查阅官方文档得出结论,redis的scan操作需要full iteration,即最优方式是一个连接将以此scan任务执行完全后释放该连接。

redis-scan-doc

修改代码如下:

RedisConnectionFactory factory = redisTemplate.getConnectionFactory();
RedisConnection rc = factory.getConnection();
if (rc instanceof JedisConnection) {
    JedisConnection JedisConnection = (JedisConnection) rc;
    Cursor<Map.Entry<String, String>> cursorResult = new ConvertingCursor<Map.Entry<byte[], byte[]>, Map.Entry<String, String>>(
            JedisConnection.hScan(rawValue(key), cursor, scanOptions),
            new Converter<Map.Entry<byte[], byte[]>, Map.Entry<String, String>>() {
            ...
            });
return new ScanResult<Map.Entry<String, String>>(cursorResult, factory, rc);}


public void releaseConnection() throws IOException{
    IOException ex = null;
    if(cursor != null) {
        try {
            cursor.close();
        } catch (IOException e) {
            ex = e;
        }
    }
    try {
        RedisConnectionUtils.releaseConnection(rc, factory);
    } catch (Exception e) {

    }
    if(ex != null) {
        throw ex;
    }
}

将连接返回给业务代码,并在业务代码执行完毕后将连接释放,问题解决。

总结

  1. 连接一旦开启就必须释放,否则造成内存泄漏或服务堵塞不可用
  2. 重写代码时需要谨记仔细查阅官方文档给出的方案并实施
  3. 多线程下使用redis的scan操作需要使用一个连接遍历完Cursor,而不能复用连接,否则导致报错Unknown reply.
上一篇下一篇

猜你喜欢

热点阅读