Member since
08-16-2016
642
Posts
131
Kudos Received
68
Solutions
My Accepted Solutions
Title | Views | Posted |
---|---|---|
3099 | 10-13-2017 09:42 PM | |
5280 | 09-14-2017 11:15 AM | |
2754 | 09-13-2017 10:35 PM | |
4354 | 09-13-2017 10:25 PM | |
4959 | 09-13-2017 10:05 PM |
07-06-2017
02:50 AM
I have increased the heap size. It was set to default of 256 MB which I guess was causing the problem. I will revert if it keeps working alright 🙂 Thanks much for your response .It helped
... View more
07-05-2017
02:11 AM
Yes, install web browser on the same maching and try accessing it from the browser. Do this in the first time, instead trying from your host (if you are using VirtualBox or other virtualization tools). You can ignore the "unable to retrieve non-local non-loopback ip address" error.
... View more
07-02-2017
07:57 PM
I have been re-run the test, and kudu perform much better this time(though it's still a little bit slower than parquet), thanks for @mpercy's suggestion. I changed two things by re-runing the test: 1, increase the partitions for the fact table from 60 to 768(affact all queries) 2, change the query3.sql 'or' predicate into 'in' predicate, so predicate can push down to kudu(only affact query 3) below is the re-run result: (column 'kudu60' is the previous result, which means the partitions of fact table is 60 ) (column 'kudu768' is the new result, which means the partitions of fact table is 768)
... View more
06-30-2017
06:34 AM
Interesting story. The decomm process would not complete until all blocks have at least 1 good replica on other DNs. (good replica = replicas that are not stale and on a DataNode that is not being decommissioned or already decommissioned) DirectoryScanner in a DataNode scans the entire directory, reconciling inconsistency between in-memory block map and on-disk replica, so it would eventually pick up the added replica, just a matter of time.
... View more
06-29-2017
06:43 PM
This usually means that another adaptor is being picked up by the test. For me it was the loopback and it doesn't have a speed or mode, so the health test fails. Use ethtool to examine your adaptors and find the one that doesn't have a speed. Add a regex to exclude it in Network Interface Collection Exclusion Regex under the Host configuration screen. My regex for the loopback adaptor (lo) is ^lo$
... View more
06-28-2017
08:53 AM
1 Kudo
Thank you very much @mbigelow I was able to fix the missing blocks issue & replication factor change. Missing block issue : All the datanodes were included in rack but it's the configuration issue we had in our cluster that caused the issue. Replication factor : Yes we need to change the client configuration value and re-deployed client configuration files and restarted the HDFS , YARN and all other client services that require this update. Following are the links were useful for me to change client configuration files https://www.cloudera.com/documentation/enterprise/5-6-x/topics/cm_mc_mod_configs.html http://grokbase.com/t/cloudera/scm-users/126wjwf5da/setting-the-replication-value
... View more
06-23-2017
06:30 AM
This has to do with the YARN memory settings. The amount of memory allocated to yarn is only 8 GB. I don't know what the minimum container size is, probably around 1.3 G. That combination of the two determine the amount of containers that can be launched. The result of that for your cluster is 6 containers. Anything beyond that will have to wait for resources to be freed up. https://hortonworks.com/blog/how-to-plan-and-configure-yarn-in-hdp-2-0/ https://www.cloudera.com/documentation/enterprise/5-3-x/topics/cdh_ig_yarn_tuning.html
... View more
06-23-2017
02:54 AM
1 Kudo
In the logs for the ApplicationMaster/SparkDriver (which was around 4GB) I've found a StackOverflowError from Spark reporter thread: I've found this Spark issue https://issues.apache.org/jira/browse/SPARK-18750 that matches my error. The job was launched used dynamicAllocation and requested an insane number of containers (16000 with 20GB/8cores) and apparently this can cause a SO in the Spark thread managing the executors. An easy workaround is to disable dynamicAllocation and use a fixed number of executor. With 10 executors the job is running fine.
... View more
06-20-2017
08:40 PM
Partition column is a logical entity related to the table. It cannot be present in the data or schema of the Hive table. So how can I put the data in DataFrame and a case class ?
... View more