[
https://issues.apache.org/jira/browse/SPARK-15678?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15877513#comment-15877513
]
Gen TANG edited comment on SPARK-15678 at 2/22/17 5:41 AM:
-----------------------------------------------------------
hi, I found a bug which is probably related with this issue.[~sameerag]
Please consider the following code.
{code}
import org.apache.spark.sql.DataFrame
def f(data: DataFrame): DataFrame = {
val df = data.filter("id>10")
df.cache
df.count
df
}
f(spark.range(100).asInstanceOf[DataFrame]).count // output 89 which is correct
f(spark.range(1000).asInstanceOf[DataFrame]).count // output 989 which is
correct
val dir = "/tmp/test"
spark.range(100).write.mode("overwrite").parquet(dir)
val df = spark.read.parquet(dir)
df.count // output 100 which is correct
f(df).count // output 89 which is correct
spark.range(1000).write.mode("overwrite").parquet(dir)
val df1 = spark.read.parquet(dir)
df1.count // output 1000 which is correct, in fact other operation expect
df1.filter("id>10") return correct result.
f(df1).count // output 89 which is incorrect
{code}
In fact, when we use df1.filter("id>10"), spark will use old cached dataFrame.
was (Author: gen):
hi, I found a bug which is probably related with this issue.[~sameerag]
Please consider the following code.
{code}
import org.apache.spark.sql.DataFrame
def f(data: DataFrame): DataFrame = {
val df = data.filter("id>10")
df.cache
df.count
df
}
f(spark.range(100).asInstanceOf[DataFrame]).count // output 89 which is correct
f(spark.range(1000).asInstanceOf[DataFrame]).count // output 989 which is
correct
val dir = "/tmp/test"
spark.range(100).write.mode("overwrite").parquet(dir)
val df = spark.read.parquet(dir)
df.count // output 100 which is correct
f(df).count // output 89 which is correct
spark.range(1000).write.mode("overwrite").parquet(dir)
val df1 = spark.read.parquet(dir)
df1.count // output 1000 which is correct, in fact other operation expect
df1.filter("id>10") return correct result.
f(df1).count // output 89 which is incorrect
{code}
> Not use cache on appends and overwrites
> ---------------------------------------
>
> Key: SPARK-15678
> URL: https://issues.apache.org/jira/browse/SPARK-15678
> Project: Spark
> Issue Type: Bug
> Affects Versions: 2.0.0
> Reporter: Sameer Agarwal
> Assignee: Sameer Agarwal
> Fix For: 2.0.0
>
>
> SparkSQL currently doesn't drop caches if the underlying data is overwritten.
> {code}
> val dir = "/tmp/test"
> sqlContext.range(1000).write.mode("overwrite").parquet(dir)
> val df = sqlContext.read.parquet(dir).cache()
> df.count() // outputs 1000
> sqlContext.range(10).write.mode("overwrite").parquet(dir)
> sqlContext.read.parquet(dir).count() // outputs 1000 instead of 10 <---- We
> are still using the cached dataset
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
