- android - RelativeLayout 背景可绘制重叠内容
- android - 如何链接 cpufeatures lib 以获取 native android 库?
- java - OnItemClickListener 不起作用,但 OnLongItemClickListener 在自定义 ListView 中起作用
- java - Android 文件转字符串
我是 Hadoop 的新手,尝试使用 Hadoop 编写关系连接。该算法尝试在连续两轮中连接三个关系。我使用递归方法。该程序运行良好。但是在执行期间它会尝试打印这样的警告:
14/12/02 10:41:16 WARN io.ReadaheadPool: Failed readahead on ifile
EBADF: Bad file descriptor
at org.apache.hadoop.io.nativeio.NativeIO$POSIX.posix_fadvise(Native Method)
at org.apache.hadoop.io.nativeio.NativeIO$POSIX.posixFadviseIfPossible(NativeIO.java:263)
at org.apache.hadoop.io.nativeio.NativeIO$POSIX$CacheManipulator.posixFadviseIfPossible(NativeIO.java:142)
at org.apache.hadoop.io.ReadaheadPool$ReadaheadRequestImpl.run(ReadaheadPool.java:206)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)
at java.lang.Thread.run(Thread.java:745)
这很烦人,我想知道问题的原因以及如何摆脱它们。我的代码如下:
public class Recursive {
/**
* Join three relations together using recursive method
* R JOIN S JOIN T = ((R JOIN S) JOIN T)
*/
static String[] relationSequence; // Keeps sequence of relations in join
static int round; // Round number running
/**
* Mapper
* Relation name = R
* Input tuple = a b
* Output pair = (b, (R,a))
* We assume that join value is the last attribute for the first relation
* and the first attribute for the second relation.
* So using this assumption, this map-reduce algorithm will work for any number of attributes
*/
public static class joinMapper extends Mapper<Object, Text, IntWritable, Text>{
public void map(Object keyIn, Text valueIn, Context context) throws IOException, InterruptedException {
// Read tuple and put attributes in a string array
String curValue = valueIn.toString();
String[] values = curValue.split("\t");
// Get relation name from input file name
String fileName = ((FileSplit) context.getInputSplit()).getPath().getName();
// Get join attribute index number R join S
int joinIndex;
String others = "";
if(fileName.compareTo(relationSequence[round])==0){
joinIndex = 0;
others = curValue.substring(0+2);
}else{
joinIndex = values.length - 1;
others = curValue.substring(0, curValue.length()-2);
}
IntWritable joinValue = new IntWritable(Integer.parseInt(values[joinIndex]));
// Create list of attributes which are not join attribute
Text temp = new Text(fileName + "|" + others);
context.write(joinValue,temp);
}
}
/**
* Reducer
*
* 1. Divide the input list in two ArrayLists based on relation name:
* a. first relation
* b. second relation
* 2. Test if the second relation is not empty. If it's so, we shouldn't continue.
* 3. For each element of the first array list, join it with the all elements in
* the second array list
*/
public static class joinReducer extends Reducer<IntWritable, Text, Text, Text>{
public void reduce(IntWritable keyIn, Iterable<Text> valueIn, Context context)
throws IOException, InterruptedException{
ArrayList<String> firstRelation = new ArrayList<String>();
ArrayList<String> secondRelation = new ArrayList<String>();
for (Text value : valueIn) {
String[] values = value.toString().split("\\|");
if(values[0].compareTo(relationSequence[round])==0){
secondRelation.add(values[1]);
}else{
firstRelation.add(values[1]);
}
}
if(secondRelation.size()>0){
for (String firstItem : firstRelation) {
for (String secondItem : secondRelation) {
context.write(new Text(firstItem.toString()), new Text(keyIn.toString() + "\t"
+ secondItem.toString()
));
}
}
}
}
}
/**
* Partitioner
*
* In order to hash pairs to reducer tasks, we used logical which is
* obviously faster than module function.
*/
public static class joinPartitioner extends Partitioner<IntWritable, Text> {
public int getPartition(IntWritable key, Text value, int numReduceTasks) {
int partitionNumber = key.get()&0x007F;
return partitionNumber;
}
}
/**
* Main method
*
* (R join S join T)
* hadoop jar ~/COMP6521.jar Recursive /input/R /input/S /input2/T /output R,S,T
*
* @param args
* <br> args[0]: first relation
* <br> args[1]: second relation
* <br> args[2]: third relation
* <br> args[3]: output directory
* <br> args[4]: relation sequence to join, separated by comma
*/
public static void main(String[] args) throws IllegalArgumentException, IOException, InterruptedException, ClassNotFoundException {
long s = System.currentTimeMillis();
/****** Preparing problem variables *******/
relationSequence = args[4].split(","); // Keep sequence of relations
round = 1; // Variable to keep current round number
int maxOfReducers = 128; // Maximum number of available reducers
int noReducers; // Number of reducers for one particular job
noReducers = maxOfReducers;
Path firstRelation = new Path(args[0]);
Path secondRelation = new Path(args[1]);
Path thirdRelation = new Path(args[2]);
Path temp = new Path("/temp"); // Temporary path to keep intermediate result
Path out = new Path(args[3]);
/****** End of variable Preparing *******/
Configuration conf = new Configuration();
/****** Configuring first job *******/
// General configuration
Job job = Job.getInstance(conf, "Recursive multi-way join (first round)");
job.setNumReduceTasks(noReducers);
// Pass appropriate classes
job.setJarByClass(Recursive.class);
job.setMapperClass(joinMapper.class);
job.setPartitionerClass(joinPartitioner.class);
job.setReducerClass(joinReducer.class);
// Specify input and output type of reducers
job.setOutputKeyClass(Text.class);
job.setOutputValueClass(Text.class);
job.setMapOutputKeyClass(IntWritable.class);
job.setMapOutputValueClass(Text.class);
FileSystem fs = FileSystem.get(conf);
if(fs.exists(temp)){ fs.delete(temp, true);}
if(fs.exists(out)) { fs.delete(out, true); }
// Specify the input and output paths
FileInputFormat.addInputPath(job, firstRelation);
FileInputFormat.addInputPath(job, secondRelation);
FileOutputFormat.setOutputPath(job, temp);
/****** End of first job configuration *******/
job.submit();
// Running the first job
boolean b = job.waitForCompletion(true);
if(b){
// try to execute the second job after completion of the first one
round++; // Specify round number
Configuration conf2 = new Configuration(); // Create new configuration object
/****** Configuring second job *******/
// General configuration
Job job2 = Job.getInstance(conf2, "Reduce multi-way join (second round)");
job2.setNumReduceTasks(noReducers);
// Pass appropriate classes
job2.setJarByClass(Recursive.class);
job2.setMapperClass(joinMapper.class);
job2.setPartitionerClass(joinPartitioner.class);
job2.setReducerClass(joinReducer.class);
// Specify input and output type of reducers
job2.setOutputKeyClass(Text.class);
job2.setOutputValueClass(Text.class);
// Specify input and output type of mappers
job2.setMapOutputKeyClass(IntWritable.class);
job2.setMapOutputValueClass(Text.class);
// End of 2014-11-25
// Specify the input and output paths
FileInputFormat.addInputPath(job2, temp);
FileInputFormat.addInputPath(job2, thirdRelation);
FileOutputFormat.setOutputPath(job2, out);
/****** End of second job configuration *******/
job2.submit();
// Running the first job
b = job2.waitForCompletion(true);
// Output time measurement
long e = System.currentTimeMillis() - s;
System.out.println("Total: " + e);
System.exit(b ? 0 : 1);
}
System.exit(1);
}
}
最佳答案
我有一个类似的错误,我结束了你的问题,这个 mail list thread EBADF: Bad file descriptor
To clarify a little bit, the readahead pool can sometimes spit out this message if you close a file while a readahead request is in flight. It's not an error and just reflects the fact that the file was closed hastily, probably because of some other bug which is the real problem.
在我的例子中,我关闭了一个 writer 而没有用 hflush
由于您似乎没有手动使用 writer 或 reader,我可能会看看您是如何发送 mr 任务的。
关于hadoop 警告 EBADF : Bad file descriptor,我们在Stack Overflow上找到一个类似的问题: https://stackoverflow.com/questions/27253616/
你好,我正在尝试在 opensuse 中创建一个 Shell 脚本来创建 MySqlUsers,但是当我尝试运行它时,我得到了这个错误: Warning: Could not start progra
我阅读了有关此错误的所有信息,但未能找到任何解决方案。 我有一个看起来像这样的简单页面: $xmlfile = "/var/www/marees.xml"; //Fichier dans lequel
运行 Websphere App 服务器 V8.5 Liberty Profile。我找不到任何可以解决这些警告的帮助。我在 eclipse 。 ************** He
我尝试在 GC AppEngine 上部署应用程序。部署过程中没有错误,但应用程序无法运行(仅显示加载页面)。日志中唯一一个奇怪的原始 OpenBLAS WARNING - could not det
我刚开始学习 RestKit。我正在尝试使用它来使用 Foursquare api 获取附近的 field 。但每次我尝试“objectLoader:(RKObjectLoader *)objectL
我对 Vuejs 比较陌生,每次按键时都会收到以下警告: [Vue warn]: $attrs is readonly. found in ---> at src\component
Warning: simplexml_load_file() [function.simplexml-load-file]: I/O warning : failed to load external
我在尝试修改某些表时不断收到此错误。这是我的代码: /** = 1){ //$this->mysqli->autocommit(FALSE); //insert th
当我尝试使用 PHP 的 ftp_put 函数上传文件时,早些时候出现错误: 警告:ftp_put() [function.ftp-put]:无数据连接 现在,我尝试开启被动模式: ftp_pasv(
我一直在努力让这段代码适用于现阶段的年龄。它旨在计算一个范围内的素数,我已经编写了一种方法来打印它们。不幸的是,代码将无法编译,引用警告: “警告:[未检查] 未检查调用 add(E) 作为原始类型
尝试使用带有架构组件和Kotlin的Android Studio 3 Canary 5构建示例会给出此警告。 谁能告诉我原因? 谢谢,Ove 编辑#1: 这是Dan Lew前段时间制作的样本 http
我正在编写一个 Shiny 的应用程序,它运行得非常好,突然我收到两条警告消息。我已经回到以前运行良好的副本,它们现在显示相同的错误消息,所以我真的很困惑。我的代码仍然运行并在我 Shiny 的仪表板
03-25 05:52:15.329 8029-8042/com.mgh.radio W/MediaPlayerNative: info/warning (703, 0) 03-25 05:52:15
我在构建时在我的 gradle 控制台中收到一条警告消息: 警告:[options] 引导类路径未与 -source 1.7 一起设置 1 条警告 我怎样才能解决这个问题? 任何帮助表示赞赏! 最佳答
我有下一个代码: 测试.c #include "a1.h" int main() { int a = 8; foo(a); return a; } a1.h void foo
我的程序中有一个 WORD 变量。 WORD hour; 但是当我比较它的时候 if(hour>=0 && hour=0 && hour=0 的比较,它始终适用于 hour 是 WORD 类型,它是一
安全研究人员警告称,一个最新的严重的Java错误,其本质与目前在全球范围内利用的臭名昭著的 Log4Shell 漏洞相同 。 CVE-2021-42392 尚未在国家漏洞数据库 (NVD) 中
安装SqlServer2005时“版本变更检查 (警告)"问题排查 今天同事在安装SqlServer2005时遇到“版本变更检查 (警告) ”问题导致安装失败,警告提示如下: - 版本
我的 UWP 项目中出现以下警告。我已经标记了解决方案的示例,但我更感兴趣的是为什么在同一平台上创建另一个空项目时不会出现此警告? APPX4001: Build property AppxBundl
我试图修复我的登录脚本,在我的本地主机上它可以工作,但上传到我的在线测试服务器时,注销被破坏,我得到这个错误: Warning: session_destroy() [function.session
我是一名优秀的程序员,十分优秀!