chore(java-devel): Tweak Maven transport settings [skip ci] (#1584)

* chore(java-devel): Lower Maven connection TTL to 25s

* chore(java-devel): Tweak Maven retry settings

* chore(java-devel): Add potential future Maven setting

* chore(java-devel): Add link to Maven resolver settings

* chore(java-devel): Remove "l" from numbers (longs)

* chore(java-devel): Fix typo

* chore: Apply suggestions

Co-authored-by: Sebastian Bernauer <sebastian.bernauer@stackable.tech>

---------

Co-authored-by: Sebastian Bernauer <sebastian.bernauer@stackable.tech>

Author: Techassi

java-devel/stackable/settings.xml

@@ -40,6 +40,8 @@
 
       <properties>
         <!--
+        Resolver specific configuration keys can be found here: https://maven.apache.org/resolver/configuration.html
+
         This is to work around an issue where Maven builds in Github Actions would randomly fail.
         There is some evidence that points at Azure networking as the root cause where it closes idle connections (silently) after 4 minutes.
         Maven would then sometimes reuse an "old" connection from its connection pool and would encounter issues due to these connection closures.
@@ -51,10 +53,35 @@
         -->
 
         <!-- Old name: maven-resolver 1.x, Maven 3.x -->
-        <aether.connector.http.connectionMaxTtl>30</aether.connector.http.connectionMaxTtl>
+        <aether.connector.http.connectionMaxTtl>25</aether.connector.http.connectionMaxTtl>
 
         <!-- New name: maven-resolver 2.x, as of Maven 4.x -->
-        <aether.transport.http.connectionMaxTtl>30</aether.transport.http.connectionMaxTtl>
+        <aether.transport.http.connectionMaxTtl>25</aether.transport.http.connectionMaxTtl>
+
+        <!--
+        By default, requests are retried 3 times. We increase the count slightly to hopefully get less failing
+        builds in CI.
+        -->
+        <aether.transport.http.retryHandler.count>5</aether.transport.http.retryHandler.count>
+
+        <!--
+        By default, retries wait for 5000 milliseconds (5 seconds). We increase this to 10000 milliseconds
+        (10 seconds) to spam the Nexus instance a little less.
+        -->
+        <aether.transport.http.retryHandler.interval>10000</aether.transport.http.retryHandler.interval>
+
+        <!--
+        Servers can respond with `Retry-After` headers. By default, the value is capped at 300000 milliseconds (300
+        seconds). We reduce this to 60000 (60 seconds) to avoid super long retry intervals.
+        -->
+        <aether.transport.http.retryHandler.intervalMax>60000</aether.transport.http.retryHandler.intervalMax>
+
+        <!--
+        This is an option we can experiment with in the future. Settings this to false will result in no connection
+        pooling, which should help to avoid stale/long-lasting connections which can become inactive and be killed by
+        the Azure networking.
+        -->
+        <!-- <aether.transport.http.reuseConnections>false</aether.transport.http.reuseConnections> -->
       </properties>
     </profile>
   </profiles>