Member since
04-04-2016
166
Posts
168
Kudos Received
29
Solutions
My Accepted Solutions
Title | Views | Posted |
---|---|---|
2897 | 01-04-2018 01:37 PM | |
4903 | 08-01-2017 05:06 PM | |
1569 | 07-26-2017 01:04 AM | |
8917 | 07-21-2017 08:59 PM | |
2607 | 07-20-2017 08:59 PM |
01-04-2018
01:37 PM
@Zack Riesland You can create a separate table with only the current day's partition. Merge/consolidate the small files into it. Then run an exchange partition into the main table to achieve this. In that way, you do not need to play with the entire data in the main table and this kind of gives you a clean way to achieve this. Create a shell script and put the commands and do the boundary checks like the end of the day, reprocessing etc and you will have an airtight solution. Thanks, Raj
... View more
08-01-2017
05:06 PM
@Sonu Sahi and @Greg Keys There is a bug in Hortonworks > ODBC 2.1.5 This is fixed and will be part of HDP 3.0 As an interim, we can use ODBC driver v2.1.2 from the archive section under HDP 2.4. Thanks
... View more
07-28-2017
09:45 PM
++ non-LLAP works fine with the same setup just different port. Will keep posted.
... View more
07-28-2017
09:42 PM
Am trying to use http not sasl.
... View more
07-28-2017
09:40 PM
yes Sir am sure.
... View more
07-28-2017
05:51 PM
Have you managed to connect hive LLAP via ODBC using http and uid /pwd? If yes, can you provide the option you laid on the http path in the http properties? Using 'cliservice' passes the connection test but it eventually fails with the below error when using the odbc in power bi since LLAp does not support cli. DataSource.Error: ODBC: ERROR [HY000][Hortonworks][Hardy] (35)Errorfrom server: error code:'0' error message: 'MetaException(message:java.lang.NullPointerException)'. Details: DataSourceKind=Odbc DataSourcePath=dsn=Hive_DSN OdbcErrors=Table Connecting to hiveserver2 works in the above-mentioned way Thanks
... View more
Labels:
- Labels:
-
Apache Hive
07-26-2017
01:04 AM
@Manish Kumar Yadav You probably are looking at Hadoop block size vs split size. Below is a nice read: https://hadoopjournal.wordpress.com/2015/06/30/mapreduce-input-split-versus-hdfs-blocks/
... View more
07-26-2017
12:34 AM
@Harish Nerella You can use usual hive update to update bulk records: https://cwiki.apache.org/confluence/display/Hive/LanguageManual+DML#LanguageManualDML-Update Merge is more flexible and a bit different than a plain update. But if you are only performing only updates I do not see any reason not be able to do it using update statement from Hive 0.14.
... View more
07-21-2017
08:59 PM
1 Kudo
@PPR Reddy Here goes the solution (You can do it in other ways if you choose to): Column name in table a is c hive> select * from a; OK y y y n hive> Query: hive> select c,per from( > select x.c c,(x.cc/y.ct)*100 per,(z.cn/y.ct)*100 pern from > (select c, count(*) cc from a group by c) x, (select count(*)ct from a) y, > (select c, count(*) cn from a where c='n' group by c) z) f > where pern > 20; Output: OK n 25.0 y 75.0 Thanks
... View more
07-21-2017
04:34 PM
@Laurent lau that equal distribution of replica is not guaranteed. If you think about it in high level it does compromise the speed of writes as well as reads. So not recommended even if you are planning to do it programmatically.
... View more