|
|
-
Getting CNF during map reduce execution
buddhika chamith 2012-04-05, 17:12
Hi All,
I am trying to use hive inside an OSGi enviornment. So far I faced several OSGi issues in hive and some issues related to JDO implementation used (datanucleaus) under the OSGi enviornemnt and was able to overcome them. However when I run a query involving a map reduce operation I am getting a ClassNotFound exception from hadoop side. (Hadoop is running in pseudo distributed mode in the local machine).
java.lang.RuntimeException: java.lang.RuntimeException: java.lang.ClassNotFoundException: org.apache.hadoop.hive.shims.HadoopShimsSecure$NullOutputCommitter at org.apache.hadoop.conf.Configuration.getClass(Configuration.java:898) at org.apache.hadoop.mapred.JobConf.getOutputCommitter(JobConf.java:613) at org.apache.hadoop.mapred.Task.initialize(Task.java:492) at org.apache.hadoop.mapred.MapTask.run(MapTask.java:352) I think this is probably due to some jar/ resource not being bundled properly in to the deployed archive to Hadoop. Would some one be able to explain a bit of how the job submission part from Hive to Hadoop works and any possible reason for this behaviour? Any pointers to look in to the code is also appreciated.
Regards Buddhika
+
buddhika chamith 2012-04-05, 17:12
-
Re: Getting CNF during map reduce execution
Carl Steinbach 2012-04-05, 19:19
HI Buddhika,
Which version of Hive are you running?
Thanks.
Carl
On Thu, Apr 5, 2012 at 10:12 AM, buddhika chamith <[EMAIL PROTECTED]>wrote:
> Hi All, > > I am trying to use hive inside an OSGi enviornment. So far I faced several > OSGi issues in hive and some issues related to JDO implementation used > (datanucleaus) under the OSGi enviornemnt and was able to overcome them. > However when I run a query involving a map reduce operation I am getting a > ClassNotFound exception from hadoop side. (Hadoop is running in pseudo > distributed mode in the local machine). > > java.lang.RuntimeException: java.lang.RuntimeException: java.lang.ClassNotFoundException: org.apache.hadoop.hive.shims.HadoopShimsSecure$NullOutputCommitter > at org.apache.hadoop.conf.Configuration.getClass(Configuration.java:898) > at org.apache.hadoop.mapred.JobConf.getOutputCommitter(JobConf.java:613) > at org.apache.hadoop.mapred.Task.initialize(Task.java:492) > at org.apache.hadoop.mapred.MapTask.run(MapTask.java:352) > > > I think this is probably due to some jar/ resource not being bundled > properly in to the deployed archive to Hadoop. Would some one be able to > explain a bit of how the job submission part from Hive to Hadoop works and > any possible reason for this behaviour? Any pointers to look in to the code > is also appreciated. > > Regards > Buddhika >
+
Carl Steinbach 2012-04-05, 19:19
-
Re: Getting CNF during map reduce execution
shashwat shriparv 2012-04-05, 19:21
Your hive jars in hadoop and restart the hadoop then try.
On Fri, Apr 6, 2012 at 12:49 AM, Carl Steinbach <[EMAIL PROTECTED]> wrote:
> HI Buddhika, > > Which version of Hive are you running? > > Thanks. > > Carl > > On Thu, Apr 5, 2012 at 10:12 AM, buddhika chamith <[EMAIL PROTECTED] > > wrote: > >> Hi All, >> >> I am trying to use hive inside an OSGi enviornment. So far I faced >> several OSGi issues in hive and some issues related to JDO implementation >> used (datanucleaus) under the OSGi enviornemnt and was able to overcome >> them. However when I run a query involving a map reduce operation I am >> getting a ClassNotFound exception from hadoop side. (Hadoop is running in >> pseudo distributed mode in the local machine). >> >> java.lang.RuntimeException: java.lang.RuntimeException: java.lang.ClassNotFoundException: org.apache.hadoop.hive.shims.HadoopShimsSecure$NullOutputCommitter >> at org.apache.hadoop.conf.Configuration.getClass(Configuration.java:898) >> at org.apache.hadoop.mapred.JobConf.getOutputCommitter(JobConf.java:613) >> at org.apache.hadoop.mapred.Task.initialize(Task.java:492) >> at org.apache.hadoop.mapred.MapTask.run(MapTask.java:352) >> >> >> I think this is probably due to some jar/ resource not being bundled >> properly in to the deployed archive to Hadoop. Would some one be able to >> explain a bit of how the job submission part from Hive to Hadoop works and >> any possible reason for this behaviour? Any pointers to look in to the code >> is also appreciated. >> >> Regards >> Buddhika >> > > -- ∞ Shashwat Shriparv
+
shashwat shriparv 2012-04-05, 19:21
-
Re: Getting CNF during map reduce execution
buddhika chamith 2012-04-09, 08:47
Hi,
Sorry I forgot to mention the versions. I am running Hive 0.8.1 and Hadoop 0.20.203.0. Shashwat's suggestion worked :). (I dropped hive-exec to hadoop lib). Thanks. Anyway I am curious how this happened since with hive running standalone (i mean not inside OSGi) this issue didn't come up. May be hive-exec not getting uploaded to hdfs properly during hadoop job submission?
Regards Buddhika On Fri, Apr 6, 2012 at 1:21 AM, shashwat shriparv <[EMAIL PROTECTED] > wrote:
> Your hive jars in hadoop and restart the hadoop then try. > > > On Fri, Apr 6, 2012 at 12:49 AM, Carl Steinbach <[EMAIL PROTECTED]> wrote: > >> HI Buddhika, >> >> Which version of Hive are you running? >> >> Thanks. >> >> Carl >> >> On Thu, Apr 5, 2012 at 10:12 AM, buddhika chamith < >> [EMAIL PROTECTED]> wrote: >> >>> Hi All, >>> >>> I am trying to use hive inside an OSGi enviornment. So far I faced >>> several OSGi issues in hive and some issues related to JDO implementation >>> used (datanucleaus) under the OSGi enviornemnt and was able to overcome >>> them. However when I run a query involving a map reduce operation I am >>> getting a ClassNotFound exception from hadoop side. (Hadoop is running in >>> pseudo distributed mode in the local machine). >>> >>> java.lang.RuntimeException: java.lang.RuntimeException: java.lang.ClassNotFoundException: org.apache.hadoop.hive.shims.HadoopShimsSecure$NullOutputCommitter >>> at org.apache.hadoop.conf.Configuration.getClass(Configuration.java:898) >>> at org.apache.hadoop.mapred.JobConf.getOutputCommitter(JobConf.java:613) >>> at org.apache.hadoop.mapred.Task.initialize(Task.java:492) >>> at org.apache.hadoop.mapred.MapTask.run(MapTask.java:352) >>> >>> >>> I think this is probably due to some jar/ resource not being bundled >>> properly in to the deployed archive to Hadoop. Would some one be able to >>> explain a bit of how the job submission part from Hive to Hadoop works and >>> any possible reason for this behaviour? Any pointers to look in to the code >>> is also appreciated. >>> >>> Regards >>> Buddhika >>> >> >> > > > -- > > > ∞ > Shashwat Shriparv > > >
+
buddhika chamith 2012-04-09, 08:47
|
|