Update bigdata-test dependency to 0.1.6 and introduce dns.hosts for local container network alias mapping

Author: CoderYellow

README.md

@@ -50,11 +50,13 @@ GRADLE_USER_HOME=/data/.gradle ./gradlew :example:spark:dependencies \
   bundled Hadoop native libraries and prepends `-Djava.library.path` / `-Dhadoop.home.dir` to the
   Surefire `argLine`, so the native Hadoop code is used instead of the pure-Java fallback.
 - **[java-dns](../java-dns) `0.1.1`** — the agent jar (`org.openprojectx.java.dns:core`) is attached
-  to the Surefire JVM via `-javaagent:…=hosts=fake-gcs/127.0.0.1`. This redirects the `fake-gcs`
-  host to the local container, so the GCS connector can use the stable URL
-  `http://fake-gcs:4443/` and the test **no longer needs to read the dynamic GCS endpoint**. The
-  fake-gcs container is bound to the fixed host port `4443` via `[ports] fakeGcs = 4443` in
-  `spark-bigdata-test-common.toml` so the redirected host:port lines up.
+  to the Surefire JVM via `-javaagent:…=hostsFile=${project.basedir}/dns.hosts`. The
+  [`dns.hosts`](spark-apache/dns.hosts) file maps the bigdata-test container network aliases
+  (`fake-gcs`, `hdfs`) to `127.0.0.1`, so the clients reach the local mapped containers without
+  wiring per-run endpoints into the Spark session. In particular the GCS connector can use the
+  stable URL `http://fake-gcs:4443/` and the test **no longer needs to read the dynamic GCS
+  endpoint**. The fake-gcs container is bound to the fixed host port `4443` via
+  `[ports] fakeGcs = 4443` in `spark-bigdata-test-common.toml` so the redirected host:port lines up.
 
 ## Prerequisites
 

pom.xml

@@ -24,7 +24,7 @@
         <maven.compiler.release>17</maven.compiler.release>
 
         <!-- bigdata-test framework, published locally as 0.1.1-SNAPSHOT -->
-        <bigdata-test.version>0.1.4</bigdata-test.version>
+        <bigdata-test.version>0.1.6</bigdata-test.version>
 
         <!-- shared, version-line independent dependencies -->
         <junit.version>5.11.4</junit.version>

spark-apache/dns.hosts

@@ -0,0 +1,5 @@
+# java-dns host overrides (hosts-file format: <ip> <hostname>).
+# Resolves the bigdata-test container network aliases to the local mapped containers,
+# so the test reaches them without wiring per-run endpoints into the Spark session.
+127.0.0.1 fake-gcs
+127.0.0.1 hdfs

spark-apache/pom.xml

@@ -156,7 +156,8 @@
             <scope>test</scope>
         </dependency>
 
-        <!-- java-dns agent jar (attached to the Surefire JVM to redirect the fake-gcs host) -->
+        <!-- java-dns agent jar (attached to the Surefire JVM; redirects the hosts in dns.hosts
+             (fake-gcs, hdfs) to the local mapped containers) -->
         <dependency>
             <groupId>org.openprojectx.java.dns</groupId>
             <artifactId>core</artifactId>
@@ -199,7 +200,7 @@
                 <groupId>org.apache.maven.plugins</groupId>
                 <artifactId>maven-surefire-plugin</artifactId>
                 <configuration>
-                    <argLine>@{argLine} -javaagent:${org.openprojectx.java.dns:core:jar}=hosts=fake-gcs/127.0.0.1;fallback=true ${bigdata.test.argLine}</argLine>
+                    <argLine>@{argLine} -javaagent:${org.openprojectx.java.dns:core:jar}=hostsFile=${project.basedir}/dns.hosts;fallback=true ${bigdata.test.argLine}</argLine>
                     <dependenciesToScan>
                         <dependency>org.openprojectx.bigdata.test.example:spark-test-common</dependency>
                     </dependenciesToScan>

spark-cloudera/dns.hosts

@@ -0,0 +1,5 @@
+# java-dns host overrides (hosts-file format: <ip> <hostname>).
+# Resolves the bigdata-test container network aliases to the local mapped containers,
+# so the test reaches them without wiring per-run endpoints into the Spark session.
+127.0.0.1 fake-gcs
+127.0.0.1 hdfs

spark-cloudera/pom.xml

@@ -161,7 +161,8 @@
             <scope>test</scope>
         </dependency>
 
-        <!-- java-dns agent jar (attached to the Surefire JVM to redirect the fake-gcs host) -->
+        <!-- java-dns agent jar (attached to the Surefire JVM; redirects the hosts in dns.hosts
+             (fake-gcs, hdfs) to the local mapped containers) -->
         <dependency>
             <groupId>org.openprojectx.java.dns</groupId>
             <artifactId>core</artifactId>
@@ -204,7 +205,7 @@
                 <groupId>org.apache.maven.plugins</groupId>
                 <artifactId>maven-surefire-plugin</artifactId>
                 <configuration>
-                    <argLine>@{argLine} -javaagent:${org.openprojectx.java.dns:core:jar}=hosts=fake-gcs/127.0.0.1;fallback=true ${bigdata.test.argLine}</argLine>
+                    <argLine>@{argLine} -javaagent:${org.openprojectx.java.dns:core:jar}=hostsFile=${project.basedir}/dns.hosts;fallback=true ${bigdata.test.argLine}</argLine>
                     <dependenciesToScan>
                         <dependency>org.openprojectx.bigdata.test.example:spark-test-common</dependency>
                     </dependenciesToScan>

spark-test-common/src/main/resources/spark-bigdata-test-common.toml

@@ -21,6 +21,7 @@ fakeGcs = true
 # URL http://fake-gcs:4443/ (host "fake-gcs" -> 127.0.0.1) to the running container.
 [ports]
 fakeGcs = 4443
+hdfsDataNode = 9866
 
 [containerLogs]
 mode = "FILE"