[
https://issues.apache.org/jira/browse/SPARK-31193?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
daile updated SPARK-31193:
--------------------------
Description:
I see the default value of master setting in spark-submit client
{code:java}
// Global defaults. These should be keep to minimum to avoid confusing
behavior. master = Option(master).getOrElse("local[*]")
{code}
but during our development and debugging, We will encounter this kind of problem
Exception in thread "main" org.apache.spark.SparkException: A master URL must
be set in your configuration
This conflicts with the default setting
{code:java}
//If we do
val sparkConf = new SparkConf().setAppName(“app”)
//When using the client to submit tasks to the cluster, the matser will be
overwritten by the local
sparkConf.set("spark.master", "local[*]"){code}
so we have to do like this
{code:java}
val sparkConf = new SparkConf().setAppName(“app”)
//Because the program runs to set the priority of the master, we have to first
determine whether to set the master to avoid submitting the cluster to run.
sparkConf.set("spark.master",sparkConf.get("spark.master","local[*]")){code}
so is spark.app.name
Is it better for users to handle it like submit client ?
was:
{code:java}
//代码占位符
{code}
I see the default value of master setting in spark-submit client
```scala
// Global defaults. These should be keep to minimum to avoid confusing
behavior.
master = Option(master).getOrElse("local[*]")
```
but during our development and debugging, We will encounter this kind of problem
Exception in thread "main" org.apache.spark.SparkException: A master URL must
be set in your configuration
This conflicts with the default setting
```scala
//If we do
val sparkConf = new SparkConf().setAppName(“app”)
//When using the client to submit tasks to the cluster, the matser will be
overwritten by the local
sparkConf.set("spark.master", "local[*]")
```
so we have to do like this
```scala
val sparkConf = new SparkConf().setAppName(“app”)
//Because the program runs to set the priority of the master, we have to first
determine whether to set the master to avoid submitting the cluster to run.
sparkConf.set("spark.master",sparkConf.get("spark.master","local[*]"))
```
so is spark.app.name
Is it better for users to handle it like submit client ?
> set spark.master and spark.app.name conf default value
> ------------------------------------------------------
>
> Key: SPARK-31193
> URL: https://issues.apache.org/jira/browse/SPARK-31193
> Project: Spark
> Issue Type: Improvement
> Components: Spark Core
> Affects Versions: 2.3.0, 2.3.3, 2.4.0, 2.4.2, 2.4.3, 2.4.4, 2.4.5, 3.1.0
> Reporter: daile
> Priority: Major
> Fix For: 3.1.0
>
>
> I see the default value of master setting in spark-submit client
> {code:java}
> // Global defaults. These should be keep to minimum to avoid confusing
> behavior. master = Option(master).getOrElse("local[*]")
> {code}
> but during our development and debugging, We will encounter this kind of
> problem
> Exception in thread "main" org.apache.spark.SparkException: A master URL must
> be set in your configuration
> This conflicts with the default setting
>
> {code:java}
> //If we do
> val sparkConf = new SparkConf().setAppName(“app”)
> //When using the client to submit tasks to the cluster, the matser will be
> overwritten by the local
> sparkConf.set("spark.master", "local[*]"){code}
>
> so we have to do like this
> {code:java}
> val sparkConf = new SparkConf().setAppName(“app”)
> //Because the program runs to set the priority of the master, we have to
> first determine whether to set the master to avoid submitting the cluster to
> run.
> sparkConf.set("spark.master",sparkConf.get("spark.master","local[*]")){code}
>
>
> so is spark.app.name
> Is it better for users to handle it like submit client ?
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
