Hi,

Any ideas why a staging directory would suddenly become unavailable
after the completion of the map phase but before the start of the
reduce phase?  We noticed a sporadic failure yesterday wherein all the
map tasks completed
successfully and all the reduce tasks failed.  Upon examining task
tracker logs, the following exception stack trace was revealed:

2013-01-03 02:28:17,072 WARN org.apache.hadoop.mapred.TaskTracker:
Error initializing attempt_201211150255_237458_r_000108_1:
java.io.FileNotFoundException: File does not exist:
hdfs://59.bm-hadoop.prod.nym2:54310/user/apache/.staging/job_201211150255_237458/job.xml
        at org.apache.hadoop.hdfs.DistributedFileSystem.getFileStatus(DistributedFileSystem.java:562)
        at org.apache.hadoop.fs.FileUtil.copy(FileUtil.java:207)
        at org.apache.hadoop.fs.FileUtil.copy(FileUtil.java:157)
        at org.apache.hadoop.fs.FileSystem.copyToLocalFile(FileSystem.java:1371)
        at org.apache.hadoop.fs.FileSystem.copyToLocalFile(FileSystem.java:1352)
        at org.apache.hadoop.mapred.TaskTracker.localizeJobConfFile(TaskTracker.java:1434)
        at org.apache.hadoop.mapred.TaskTracker.initializeJob(TaskTracker.java:1318)
        at org.apache.hadoop.mapred.TaskTracker.localizeJob(TaskTracker.java:1242)
        at org.apache.hadoop.mapred.TaskTracker.startNewTask(TaskTracker.java:2541)
        at org.apache.hadoop.mapred.TaskTracker$TaskLauncher.run(TaskTracker.java:2505)

This problem doesn't seem relevant to only a specific distribution,
but for completeness we are running CDH3u3.

Thanks!

stan

Hi Stan,

I'd check the NN audit logs for the file /user/apache/.staging/
job_201211150255_237458/job.xml to see when/who deleted it away, perhaps
that would give more insight.
On Sat, Jan 5, 2013 at 2:32 AM, Stan Rosenberg <[EMAIL PROTECTED]>wrote:

> Hi,
>
> Any ideas why a staging directory would suddenly become unavailable
> after the completion of the map phase but before the start of the
> reduce phase?  We noticed a sporadic failure yesterday wherein all the
> map tasks completed
> successfully and all the reduce tasks failed.  Upon examining task
> tracker logs, the following exception stack trace was revealed:
>
> 2013-01-03 02:28:17,072 WARN org.apache.hadoop.mapred.TaskTracker:
> Error initializing attempt_201211150255_237458_r_000108_1:
> java.io.FileNotFoundException: File does not exist:
>
> hdfs://59.bm-hadoop.prod.nym2:54310/user/apache/.staging/job_201211150255_237458/job.xml
>         at
> org.apache.hadoop.hdfs.DistributedFileSystem.getFileStatus(DistributedFileSystem.java:562)
>         at org.apache.hadoop.fs.FileUtil.copy(FileUtil.java:207)
>         at org.apache.hadoop.fs.FileUtil.copy(FileUtil.java:157)
>         at
> org.apache.hadoop.fs.FileSystem.copyToLocalFile(FileSystem.java:1371)
>         at
> org.apache.hadoop.fs.FileSystem.copyToLocalFile(FileSystem.java:1352)
>         at
> org.apache.hadoop.mapred.TaskTracker.localizeJobConfFile(TaskTracker.java:1434)
>         at
> org.apache.hadoop.mapred.TaskTracker.initializeJob(TaskTracker.java:1318)
>         at
> org.apache.hadoop.mapred.TaskTracker.localizeJob(TaskTracker.java:1242)
>         at
> org.apache.hadoop.mapred.TaskTracker.startNewTask(TaskTracker.java:2541)
>         at
> org.apache.hadoop.mapred.TaskTracker$TaskLauncher.run(TaskTracker.java:2505)
>
> This problem doesn't seem relevant to only a specific distribution,
> but for completeness we are running CDH3u3.
>
> Thanks!
>
> stan
>

--
Harsh J

Thanks for following up, glad to know it is resolved!

On Mon, Jan 7, 2013 at 6:42 AM, Stan Rosenberg <[EMAIL PROTECTED]> wrote:
> On Sat, Jan 5, 2013 at 2:44 AM, Harsh J <[EMAIL PROTECTED]> wrote:
>> I'd check the NN audit logs for the file
>> /user/apache/.staging/job_201211150255_237458/job.xml to see when/who
>> deleted it away, perhaps that would give more insight.
>>
>
> The audit logs led to a trail which revealed user error.  Thanks Harsh!

--
Harsh J