HADOOP-19015. Increase fs.s3a.connection.maximum to 500 to minimize risk of Timeout waiting for connection from pool. (#6372)

HADOOP-19015.  Increase fs.s3a.connection.maximum to 500 to minimize the risk of Timeout waiting for connection from the pool

Contributed By: Mukund Thakur
 Conflicts:
	hadoop-common-project/hadoop-common/src/main/resources/core-default.xml
	hadoop-tools/hadoop-aws/src/main/java/org/apache/hadoop/fs/s3a/Constants.java
	hadoop-tools/hadoop-aws/src/site/markdown/tools/hadoop-aws/performance.md

Author: mukund-thakur

hadoop-common-project/hadoop-common/src/main/resources/core-default.xml

@@ -1362,7 +1362,7 @@
 
 <property>
   <name>fs.s3a.connection.maximum</name>
-  <value>96</value>
+  <value>500</value>
   <description>Controls the maximum number of simultaneous connections to S3.
     This must be bigger than the value of fs.s3a.threads.max so as to stop
     threads being blocked waiting for new HTTPS connections.

hadoop-tools/hadoop-aws/src/main/java/org/apache/hadoop/fs/s3a/Constants.java

@@ -152,7 +152,13 @@ private Constants() {
 
   // number of simultaneous connections to s3
   public static final String MAXIMUM_CONNECTIONS = "fs.s3a.connection.maximum";
-  public static final int DEFAULT_MAXIMUM_CONNECTIONS = 96;
+
+  /**
+   * Default value for {@link #MAXIMUM_CONNECTIONS}: {@value}.
+   * Future releases are likely to increase this value.
+   * Keep in sync with the value in {@code core-default.xml}
+   */
+  public static final int DEFAULT_MAXIMUM_CONNECTIONS = 500;
 
   /**
    * Configuration option to configure expiration time of

hadoop-tools/hadoop-aws/src/site/markdown/tools/hadoop-aws/performance.md

@@ -211,19 +211,19 @@ for parallel IO (especially uploads) by setting the properties
 | property | meaning | default |
 |----------|---------|---------|
 | `fs.s3a.threads.max`| Threads in the AWS transfer manager| 10 |
-| `fs.s3a.connection.maximum`| Maximum number of HTTP connections | 10|
+| `fs.s3a.connection.maximum`| Maximum number of HTTP connections | 500 |
 
 We recommend using larger values for processes which perform
 a lot of IO: `DistCp`, Spark Workers and similar.
 
 ```xml
 <property>
   <name>fs.s3a.threads.max</name>
-  <value>20</value>
+  <value>10</value>
 </property>
 <property>
   <name>fs.s3a.connection.maximum</name>
-  <value>20</value>
+  <value>500</value>
 </property>
 ```