Spark Performance Tunning Example 1

We had a Spark Job which was taking over 3 hours to complete.

First, We found the stage which was taking time. Refer details as below -

This is simply reading file and mapping data to final save location. So, there are not much joins or calculations involved.

Second, We saw event time line, if there is any delay due to serialization/ shuffling/ scheduling. But, there was nothing.

Third, we saw total executors and tasks processed by them.

There are 5 executors each taking over 3 hours and each executing approximately 500 tasks. Which means that a task is taking almost 2.8 -3 minutes

Fourth, to confirm there is no data skew, we sorted the tasks to see maximum duration and maximum input a task processed, and maximum shuffle. And, we found nothing.

So conclusively, we could say that there is no problem with Job. Why Job is slow, is because it has less number of executors or eventually less vcores for task processing. Thus, we bumped up Number of Executors from 5 to 30. And, each executor was assigned 3 vcores. Thus, a total vcores of 30*3 = 900 was provided to Job.

This Job completed in 40 minutes which was earlier taking 3.5 hours, just by increasing number of executors.

Comments

Hive Parse JSON with Array Columns and Explode it in to Multiple rows.

Say we have a JSON String like below - { "billingCountry":"US" "orderItems":[ { "itemId":1, "product":"D1" }, { "itemId":2, "product":"D2" } ] } And, our aim is to get output parsed like below - itemId product 1 D1 2 D2 First, We can parse JSON as follows to get JSON String get_json_object(value, '$.orderItems.itemId') as itemId get_json_object(value, '$.orderItems.product') as product Second, Above will result String value like "[1,2]". We want to convert it to Array as follows - split(regexp_extract(get_json_object(value, '$.orderItems.itemId'),'^\\["(.*)\\"]$',1),'","') as itemId split(regexp_extract(get_json_object(value, '$.orderItems.product'),'^\\["(.*)\\"]$',1),...

org.apache.spark.sql.AnalysisException: Cannot overwrite a path that is also being read from.;

Caused by: org.apache.spark.sql.AnalysisException: Cannot overwrite a path that is also being read from.; at org.apache.spark.sql.execution.command.DDLUtils$.verifyNotReadPath(ddl.scala:906) at org.apache.spark.sql.execution.datasources.DataSourceAnalysis$$anonfun$apply$1.applyOrElse(DataSourceStrategy.scala:192) at org.apache.spark.sql.execution.datasources.DataSourceAnalysis$$anonfun$apply$1.applyOrElse(DataSourceStrategy.scala:134) at org.apache.spark.sql.catalyst.trees.TreeNode$$anonfun$2.apply(TreeNode.scala:267) at org.apache.spark.sql.catalyst.trees.TreeNode$$anonfun$2.apply(TreeNode.scala:267) at org.apache.spark.sql.catalyst.trees.CurrentOrigin$.withOrigin(TreeNode.scala:70) at org.apache.spark.sql.catalyst.trees.TreeNode.transformDown(TreeNode.scala:266) at org.apache.spark.sql.catalyst.trees.TreeNode.transform(TreeNode.scala:256) at org.apache.spark.sql.execution.datasources.DataSourceAnalysis.apply(DataSourceStrategy.scala:134) at org.apache.spark.sql.execution.dataso...

Read from a hive table and write back to it using spark sql

In context to Spark 2.2 - if we read from an hive table and write to same, we get following exception- scala > dy . write . mode ( "overwrite" ). insertInto ( "incremental.test2" ) org . apache . spark . sql . AnalysisException : Cannot insert overwrite into table that is also being read from .; org . apache . spark . sql . AnalysisException : Cannot insert overwrite into table that is also being read from .; 1. This error means that our process is reading from same table and writing to same table. 2. Normally, this should work as process writes to directory .hiveStaging... 3. This error occurs in case of saveAsTable method, as it overwrites entire table instead of individual partitions. 4. This error should not occur with insertInto method, as it overwrites partitions not the table. 5. A reason why this happening is because Hive table has following Spark TBLProperties in its definition. This problem will solve for insertInto met...

Hadoop Distcp Error Duplicate files in input path

One may face following error while copying data from one cluster to other, using Distcp Command: hadoop distcp -i {src} {tgt} Error: org.apache.hadoop.toolsCopyListing$DulicateFileException: File would cause duplicates. Ideally there can't be same file names. So, what might be happening in your case is you trying to copy partitioned table from one cluster to other. And, 2 different named partitions have same file name. Your solution is to correct Source path {src} in your command, such that you provide path uptil partitioned sub directory, not the file. For ex - Refer below : /a/partcol=1/file1.txt /a/partcol=2/file1.txt If you use {src} as "/a/*/*" then you will get the error "File would cause duplicates." But, if you use {src} as "/a" then you will not get error in copying.

Caused by: java.lang.UnsupportedOperationException: org.apache.parquet.column.values.dictionary.PlainValuesDictionary$PlainIntegerDictionary

Exception - Caused by: java.lang.UnsupportedOperationException: org.apache.parquet.column.values.dictionary.PlainValuesDictionary$PlainIntegerDictionary at org.apache.parquet.column.Dictionary.decodeToBinary(Dictionary.java:44) at org.apache.spark.sql.execution.vectorized.ColumnVector.getUTF8String(ColumnVector.java:645) at org.apache.spark.sql.catalyst.expressions.GeneratedClass$GeneratedIterator.processNext(Unknown Source) Analysis - This might occur because of data type mismatch between Hive Table & written Parquet file. Solution - Correct the data type to match between Hive Table & Parquet

QueryDB

Search This Blog