I don’t think you should get a hive-xml from the internet. It should have connection information about a running hive metastore - if you don’t have a hive metastore service as you are running locally (from a laptop?) then you don’t really need it. You can get spark to work with it’s own.
________________________________ From: ya <xinxi...@126.com> Sent: Friday, June 7, 2019 8:26:27 PM To: Rishikesh Gawade; felixcheun...@hotmail.com; user@spark.apache.org Subject: Spark SQL in R? Dear Felix and Richikesh and list, Thank you very much for your previous help. So far I have tried two ways to trigger Spark SQL: one is to use R with sparklyr library and SparkR library; the other way is to use SparkR shell from Spark. I am not connecting a remote spark cluster, but a local one. Both failed with or without hive-site.xml. I suspect the content of hive-site.xml I found online was not appropriate for this case, as the spark session can not be initialized after adding this hive-site.xml. My questions are: 1. Is there any example for the content of hive-site.xml for this case? 2. I used sql() function to call the Spark SQL, is this the right way to do it? ################################### ##Here is the content in the hive-site.xml:## ################################### <configuration> <property> <name>javax.jdo.option.ConnectionURL</name> <value>jdbc:mysql://192.168.76.100:3306/hive?createDatabaseIfNotExist=true</value> <description>JDBC connect string for a JDBC metastore</description> </property> <property> <name>javax.jdo.option.ConnectionDriverName</name> <value>com.mysql.jdbc.Driver</value> <description>Driver class name for a JDBC metastore</description> </property> <property> <name>javax.jdo.option.ConnectionUserName</name> <value>root</value> <description>username to use against metastore database</description> </property> <property> <name>javax.jdo.option.ConnectionPassword</name> <value>123</value> <description>password to use against metastore database</description> </property> </configuration> ################################ ##Here is the situation happened in R:## ################################ > library(sparklyr) # load sparklyr package > sc=spark_connect(master="local",spark_home="/Users/ya/Downloads/soft/spark-2.4.3-bin-hadoop2.7") > # connect sparklyr with spark > sql('create database learnsql') Error in sql("create database learnsql") : could not find function "sql" > library(SparkR) Attaching package: ‘SparkR’ The following object is masked from ‘package:sparklyr’: collect The following objects are masked from ‘package:stats’: cov, filter, lag, na.omit, predict, sd, var, window The following objects are masked from ‘package:base’: as.data.frame, colnames, colnames<-, drop, endsWith, intersect, rank, rbind, sample, startsWith, subset, summary, transform, union > sql('create database learnsql') Error in getSparkSession() : SparkSession not initialized > Sys.setenv(SPARK_HOME='/Users/ya/Downloads/soft/spark-2.4.3-bin-hadoop2.7') > sparkR.session(sparkHome=Sys.getenv('/Users/ya/Downloads/soft/spark-2.4.3-bin-hadoop2.7')) Spark not found in SPARK_HOME: Spark package found in SPARK_HOME: /Users/ya/Downloads/soft/spark-2.4.3-bin-hadoop2.7 Launching java with spark-submit command /Users/ya/Downloads/soft/spark-2.4.3-bin-hadoop2.7/bin/spark-submit sparkr-shell /var/folders/d8/7j6xswf92c3gmhwy_lrk63pm0000gn/T//Rtmpz22kK9/backend_port103d4cfcfd2c 19/06/08 11:14:57 WARN NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable Using Spark's default log4j profile: org/apache/spark/log4j-defaults.properties Setting default log level to "WARN". To adjust logging level use sc.setLogLevel(newLevel). For SparkR, use setLogLevel(newLevel). Error in handleErrors(returnStatus, conn) : …... hundreds of lines of information and mistakes here …… > sql('create database learnsql') Error in getSparkSession() : SparkSession not initialized ################################### ##Here is what happened in SparkR shell:## #################################### Error in handleErrors(returnStatus, conn) : java.lang.IllegalArgumentException: Error while instantiating 'org.apache.spark.sql.hive.HiveSessionStateBuilder': at org.apache.spark.sql.SparkSession$.org$apache$spark$sql$SparkSession$$instantiateSessionState(SparkSession.scala:1107) at org.apache.spark.sql.SparkSession$$anonfun$sessionState$2.apply(SparkSession.scala:145) at org.apache.spark.sql.SparkSession$$anonfun$sessionState$2.apply(SparkSession.scala:144) at scala.Option.getOrElse(Option.scala:121) at org.apache.spark.sql.SparkSession.sessionState$lzycompute(SparkSession.scala:144) at org.apache.spark.sql.SparkSession.sessionState(SparkSession.scala:141) at org.apache.spark.sql.api.r.SQLUtils$$anonfun$setSparkContextSessionConf$2.apply(SQLUtils.scala:80) at org.apache.spark.sql.api.r.SQLUtils$$anonfun$setSparkContextSessionConf$2.apply(SQLUtils.scala:79) at scala.collection.TraversableLike$WithFilter$$anonfun$foreach$1.apply(TraversableLike.scala:733) at scala.collection.Iterator$class.foreach(Iterator.sca > sql('create database learnsql') Error in getSparkSession() : SparkSession not initialized Thank you very much. YA 在 2019年6月8日,上午1:44,Rishikesh Gawade <rishikeshg1...@gmail.com<mailto:rishikeshg1...@gmail.com>> 写道: Hi. 1. Yes you can connect to spark via R. If you are connecting to a remote spark cluster then you'll need EITHER a spark binary along with hive-site.xml in its config direcctory on the machine running R OR livy server installed on the cluster. You can then go on to use SparklyR, which, although has almost the same functions as of SparkR, is recommended over the latter. For the first method mentioned above, use sc <- sparklyr::spark_connect(master = "yarn-client", spark_home = Sys.getenv("SPARK_HOME"), conf = spark_config()) For the second method, use sc <- sparklyr::spark_connect( master = "livyserverIP:port", method = "livy", conf = livy_config(conf = spark_config(), username = "foo", password = "bar")) 2. The reason that you're not getting the desired result could be that hive-site.xml is missing.To be able to connect to Hive from Spark-shell/Spark-submit/SparkR/SparklyR and perform sql operations, you need to have hive-site.xml in the $SPARK_HOME/conf directory. This is hive-site.xml should contain one and only one configuration which would be 'hive.metastore.uris'. 3. In case of spark-sql shell, it should work after putting the aforementioned hive-site.xml in the config directory of Spark. If it doesn't work, then please check the syntax. Regards, Rishikesh Gawade On Thu, Jun 6, 2019, 12:18 PM ya <xinxi...@126.com<mailto:xinxi...@126.com>> wrote: Dear list, I am trying to use sparksql within my R, I am having the following questions, could you give me some advice please? Thank you very much. 1. I connect my R and spark using the library sparkR, probably some of the members here also are R users? Do I understand correctly that SparkSQL can be connected and triggered via SparkR and used in R (not in sparkR shell of spark)? 2. I ran sparkR library in R, trying to create a new sql database and a table, I could not get the database and the table I want. The code looks like below: library(SparkR) Sys.setenv(SPARK_HOME='/Users/ya/Downloads/soft/spark-2.4.3-bin-hadoop2.7') sparkR.session(sparkHome=Sys.getenv('/Users/ya/Downloads/soft/spark-2.4.3-bin-hadoop2.7')) sql("create database learnsql; use learnsql") sql(" create table employee_tbl (emp_id varchar(10) not null, emp_name char(10) not null, emp_st_addr char(10) not null, emp_city char(10) not null, emp_st char(10) not null, emp_zip integer(5) not null, emp_phone integer(10) null, emp_pager integer(10) null); insert into employee_tbl values ('0001','john','yanlanjie 1','gz','jiaoqiaojun','510006','1353'); select*from employee_tbl; “) I ran the following code in spark-sql shell, I get the database learnsql, however, I still can’t get the table. spark-sql> create database learnsql;show databases; 19/06/06 14:42:36 INFO HiveMetaStore: 0: create_database: Database(name:learnsql, description:, locationUri:file:/Users/ya/spark-warehouse/learnsql.db, parameters:{}) 19/06/06 14:42:36 INFO audit: ugi=ya ip=unknown-ip-addr cmd=create_database: Database(name:learnsql, description:, locationUri:file:/Users/ya/spark-warehouse/learnsql.db, parameters:{}) Error in query: org.apache.hadoop.hive.metastore.api.AlreadyExistsException: Database learnsql already exists; spark-sql> create table employee_tbl > (emp_id varchar(10) not null, > emp_name char(10) not null, > emp_st_addr char(10) not null, > emp_city char(10) not null, > emp_st char(10) not null, > emp_zip integer(5) not null, > emp_phone integer(10) null, > emp_pager integer(10) null); Error in query: no viable alternative at input 'create table employee_tbl\n(emp_id varchar(10) not'(line 2, pos 20) == SQL == create table employee_tbl (emp_id varchar(10) not null, --------------------^^^ emp_name char(10) not null, emp_st_addr char(10) not null, emp_city char(10) not null, emp_st char(10) not null, emp_zip integer(5) not null, emp_phone integer(10) null, emp_pager integer(10) null) spark-sql> insert into employee_tbl values ('0001','john','yanlanjie 1','gz','jiaoqiaojun','510006','1353'); 19/06/06 14:43:43 INFO HiveMetaStore: 0: get_table : db=default tbl=employee_tbl 19/06/06 14:43:43 INFO audit: ugi=ya ip=unknown-ip-addr cmd=get_table : db=default tbl=employee_tbl Error in query: Table or view not found: employee_tbl; line 1 pos 0 Does sparkSQL has different coding grammar? What did I miss? Thank you very much. Best regards, YA --------------------------------------------------------------------- To unsubscribe e-mail: user-unsubscr...@spark.apache.org<mailto:user-unsubscr...@spark.apache.org>