Thank you so much, you absolutely nailed it.
There is a stupid "SPARK_HOME" env variable pointing to Spark2.4 existed on
zsh, which is the troublemaker.
Totally forgot that and didn't realize this environment variable could cause
days frustration for me.
Yong
From: Artemis User
Sent: Thursday, March 10, 2022 3:13 PM
To: user
Subject: Re: Spark 3.1 with spark AVRO
It must be some misconfiguration in your environment. Do you perhaps have a
hardwired $SPARK_HOME env variable in your shell? An easy test would be to
place the spark-avro jar file you downloaded in the jars directory of Spark and
run spark-shell again without the packages option. This will guarantee that
the jar file is on the classpath of Spark driver and executors..
On 3/10/22 1:24 PM, Yong Zhang wrote:
Hi,
I am puzzled with this issue of Spark 3.1 version to read avro file. Everything
is done on my local mac laptop so far, and I really don't know where the issue
comes from, and I googled a lot and cannot find any clue.
I am always using Spark 2.4 version, as it is really mature. But for a new
project, I want to taste Spark 3.1, which needs to read AVRO file.
To my surprise, on my local, the Spark 3.1.3 throws error when trying to read
the avro files.
* I download the Spark 3.1.2 and 3.1.3 with Hadoop2 or 3 from
https://spark.apache.org/downloads.html
* Use JDK "1.8.0_321" on the Mac
* Untar the spark 3.1.x local
* And follow https://spark.apache.org/docs/3.1.3/sql-data-sources-avro.html
Start the spark-shell in the following exactly command:
spark-3.1.3-bin-hadoop3.2/bin/spark-shell --packages
org.apache.spark:spark-avro_2.12:3.1.3
*
And I always get the following error when read the existing test AVRO files:
scala> val pageview = spark.read.format("avro").load("/Users/user/output/raw/")
java.lang.NoClassDefFoundError:
org/apache/spark/sql/execution/datasources/v2/FileDataSourceV2
I tried different version of Spark 3.x, from Spark 3.1.2 -> 3.1.3 -> 3.2.1, and
I believe they are all under Scala 2.12, and I start the spark-shell with
"--packages org.apache.spark:spark-avro_2.12:x.x.x", which x.x.x matches the
Spark version, but I got the above wired "NoClassDefFoundError" in all cases.
Meantime, download Spark2.4.8 and start spark-shell with "--packages
org.apache.spark:spark-avro_2.11:2.4.3", I can read the exactly same ARVO file
without any issue.
I am thinking it must be done wrongly on my end, but:
* I downloaded several versions of Spark and untar them directly.
* I DIDN'T have any custom "spark-env.sh/spark-default.conf" file to
include any potential jar files to mess up things
* Straight creating a spark session under spark-shell with the correct
package and try to read avro files. Nothing more.
I have to doubt there are something wrong with the Spark 3.x avro package
releases, but I know that possiblity is very low, especailly for multi
different veresions. But the class
"org/apache/spark/sql/execution/datasources/v2/FileDataSourceV2" existed under
"spark-sql_2.12-3.1.3.jar", as blow:
jar tvf spark-sql_2.12-3.1.3.jar | grep FileDataSourceV2
15436 Sun Feb 06 22:54:00 EST 2022
org/apache/spark/sql/execution/datasources/v2/FileDataSourceV2.class
So what could be wrong?
Thanks
Yong