hive-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "r7raul1984@163.com" <r7raul1...@163.com>
Subject Re: Re: hive -e run tez query error
Date Mon, 29 Jun 2015 07:34:33 GMT
BTW , my hadoop version is 2.3.0-cdh5.0.2



r7raul1984@163.com
 
From: Jeff Zhang
Date: 2015-06-29 13:23
To: user
Subject: Re: hive -e run tez query error
> But keeping that client cache disabled when running against trunk generally kills queries
all the time with occasional errors like these.

I think tez staging directory is supposed to be deleted when JVM exit (deleteOnExit), why
killing query will cause the path deleted ?



On Sat, Jun 27, 2015 at 11:48 AM, Gopal Vijayaraghavan <gopalv@apache.org> wrote:

> perhaps deleteOnExit() is set somewhere

fs.cache.disable settings from hdfs-site.xml are usually to blame for that.

Till hive-1.0, HiveServer2 used to leak filesystem objects, so the cache
was disabled.

> 2015-06-25 15:54:33,673 INFO FSNamesystem.audit: allowed=true
>ugi=lujian (auth:SIMPLE)    ip=/10.17.28.11 <http://10.17.28.11/>
>cmd=delete
>src=/user/lujian/lujian/_tez_session_dir/abb91da9-ac07-4024-a09f-8622ee1ca
>edf   dst=null    perm=null


But keeping that client cache disabled when running against trunk
generally kills queries all the time with occasional errors like these.

Cheers,
Gopal





-- 
Best Regards

Jeff Zhang
Mime
View raw message