Reputation: 18299
I'm familiar with the Cloudera architecture but not MapR. I have a RHEL vm and previously installed the MapR client software using this documentation. I am able to submit mapreduce jobs and query HDFS as expected.
I followed this documentation (after I installed the MapR yum repo) and installed sqoop like so:
yum install mapr-sqoop
If I try to sqoop in some data, or even just issue the command sqoop
, I receive the following error:
/opt/mapr/sqoop/sqoop-1.4.4/bin/configure-sqoop: line 47: /opt/mapr/bin/versions.sh: No such file or directory
Error: /opt/mapr/hadoop/hadoop- does not exist!
Please set $HADOOP_COMMON_HOME to the root of your Hadoop installation.
I have a /opt/mapr/hadoop/hadoop-0.20.2 directory. I've tried setting HADOOP_COMMON_HOME
and HADOOP_HOME
to both /opt/mapr/hadoop
and /opt/mapr/hadoop/hadoop-0.20.2
yet still receive the same error.
-- Update:
I issued a find / -name hadoop
and noted the last line which was /usr/bin/hadoop
.
I then set HADOOP_COMMON_HOME to /usr
, and was then asked to set HADOOP_MAPRED_HOME, HBASE_HOME, and HCAT_HOME, which I all set to /usr
.
This error however is still present:
/opt/mapr/sqoop/sqoop-1.4.4/bin/configure-sqoop: line 47: /opt/mapr/bin/versions.sh: No such file or directory
I opened up this file and commented out line 47. This allowed me to use the sqoop command, but the import job failed and I received a lot of Error: Unsupported major.minor version
.
Upvotes: 0
Views: 1171
Reputation: 11
There should be a patch for this if not fixed already,
Here is temp solution:
mapr-client does not give versions.sh , only mapr-core does. Simple fix is to manually copy that file from a node with mapr-core installed and customize the versions therein. sqoop then works fine.
Upvotes: 1