Member since
07-31-2013
1924
Posts
462
Kudos Received
311
Solutions
My Accepted Solutions
Title | Views | Posted |
---|---|---|
1543 | 07-09-2019 12:53 AM | |
9292 | 06-23-2019 08:37 PM | |
8050 | 06-18-2019 11:28 PM | |
8676 | 05-23-2019 08:46 PM | |
3473 | 05-20-2019 01:14 AM |
06-06-2014
11:26 AM
1 Kudo
Stopping a job may not be easily controllable, unless perhaps you are running the job in a FairScheduler pool and you deallocate all resources out of that pool (the config change can be done in dynamic fashion). This would also stop all other jobs in that pool, however. Killing a job from within a Map Task is possible, if only in insecure environments. You can use the JobClient API and call its killJob method. Since this call relies on authentication, it may not work right out in a kerberos secured runtime cluster.
... View more
05-08-2014
03:17 PM
Apache Hadoop's symlink support is not yet production ready. It has been disabled in functionality at the moment, as work on https://issues.apache.org/jira/browse/HDFS-5298 needs to complete before it can be consumed across the platform.
... View more
05-06-2014
07:49 AM
Thanks for following up with your found solution Nicolas! Indeed a bind(…) syscall would fail if the resolved address asked of it is not found among local interfaces.
... View more
05-04-2014
12:18 AM
You could do it via some of the date based built-in functions at http://www.cloudera.com/content/cloudera-content/cloudera-docs/Impala/latest/Installing-and-Using-Impala/ciiu_functions.html#datetime_functions_unique_1 I'd imagine something such as: unix_timestamp(to_date(event_end_timestamp)) - unix_timestamp(to_date(event_start_timetsamp)) Should give you an bigint return of the number of milliseconds.
... View more
05-03-2014
11:56 PM
Thanks for following up and reporting this!
... View more
05-03-2014
11:54 PM
The Hue accounts by default are not linked to Unix accounts, so you can set any password you require.
... View more
05-03-2014
09:02 AM
1 Kudo
When using the JDBC driver, the command executes on the HiveServer2 side. The file is evaluated to locally exist on the server, which is not true in your case (it exists on the local client program machine). Try instead to load the file to HDFS first, and use a HDFS URI in the LOAD DATA statement to make the server find it. Does this make sense?
... View more
05-03-2014
08:22 AM
This error is harmless and should not be causing your DataNode any issues. Can you check your DN logs instead for an actual error, if it is not starting up?
... View more
05-03-2014
07:35 AM
1 Kudo
What version of CDH are you using? The functionality appears to be a CDH5 feature: http://archive.cloudera.com/cdh5/cdh/5/oozie/DG_SLAMonitoring.html, but is not in CDH4. Also note that the SLA alerts do not email but rather just publish notifications with the contact field in them. Read more at the above link, and see http://archive.cloudera.com/cdh5/cdh/5/oozie/DG_JMSNotifications.html for JMS based SLA notifications.
... View more
05-03-2014
07:28 AM
2 Kudos
The parameters mapreduce.input.fileinputformat.split.minsize and mapreduce.input.fileinputformat.split.maxsize should work per the CDH5 (assuming you're using CDH as this is a CDH users forum) as can be seen in the code: https://github.com/cloudera/hadoop-common/blob/cdh5.0.0-release/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-core/src/main/java/org/apache/hadoop/mapreduce/lib/input/FileInputFormat.java#L61 and https://github.com/cloudera/hadoop-common/blob/cdh5.0.0-release/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-core/src/main/java/org/apache/hadoop/mapreduce/lib/input/FileInputFormat.java#L335 CDH5 is Apache Hadoop 2.3.0 with backports from Apache Hadoop trunk.
... View more