[SPARK-24895] Spark 2.4.0 Snapshot artifacts has broken metadata due to mismatched filenames

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.4.0
• Fix Version/s: 2.4.0
• Component/s: Build
• Target Version/s: 2.4.0
• Assignee: Eric Chang
• Reporter: Eric Chang
• Created: 23/Jul/18 23:39
• Updated: 12/Dec/22 18:10
• Resolved: 24/Jul/18 22:59

Description

Spark 2.4.0 has Maven build errors because artifacts uploaded to apache maven repo has mismatched filenames:

[ERROR] Failed to execute goal org.apache.maven.plugins:maven-enforcer-plugin:1.4.1:enforce (enforce-banned-dependencies) on project spark_2.4: Execution enforce-banned-dependencies of goal org.apache.maven.plugins:maven-enforcer-plugin:1.4.1:enforce failed: org.apache.maven.shared.dependency.graph.DependencyGraphBuilderException: Could not resolve following dependencies: [org.apache.spark:spark-mllib-local_2.11:jar:2.4.0-SNAPSHOT (compile), org.apache.spark:spark-network-shuffle_2.11:jar:2.4.0-SNAPSHOT (compile), org.apache.spark:spark-sketch_2.11:jar:2.4.0-SNAPSHOT (compile)]: Could not resolve dependencies for project com.databricks:spark_2.4:pom:1: The following artifacts could not be resolved: org.apache.spark:spark-mllib-local_2.11:jar:2.4.0-SNAPSHOT, org.apache.spark:spark-network-shuffle_2.11:jar:2.4.0-SNAPSHOT, org.apache.spark:spark-sketch_2.11:jar:2.4.0-SNAPSHOT: Could not find artifact org.apache.spark:spark-mllib-local_2.11:jar:2.4.0-20180723.232411-177 in apache-snapshots ([https://repository.apache.org/snapshots/]) -> [Help 1]

 

If you check the artifact metadata you will see the pom and jar files are 2.4.0-20180723.232411-177 instead of 2.4.0-20180723.232410-177:

<metadata modelVersion="1.1.0">
  <groupId>org.apache.spark</groupId>
  <artifactId>spark-mllib-local_2.11</artifactId>
  <version>2.4.0-SNAPSHOT</version>
  <versioning>
    <snapshot>
      <timestamp>20180723.232411</timestamp>
      <buildNumber>177</buildNumber>
    </snapshot>
    <lastUpdated>20180723232411</lastUpdated>
    <snapshotVersions>
      <snapshotVersion>
        <extension>jar</extension>
        <value>2.4.0-20180723.232411-177</value>
        <updated>20180723232411</updated>
      </snapshotVersion>
      <snapshotVersion>
        <extension>pom</extension>
        <value>2.4.0-20180723.232411-177</value>
        <updated>20180723232411</updated>
      </snapshotVersion>
      <snapshotVersion>
        <classifier>tests</classifier>
        <extension>jar</extension>
        <value>2.4.0-20180723.232410-177</value>
        <updated>20180723232411</updated>
      </snapshotVersion>
      <snapshotVersion>
        <classifier>sources</classifier>
        <extension>jar</extension>
        <value>2.4.0-20180723.232410-177</value>
        <updated>20180723232411</updated>
      </snapshotVersion>
      <snapshotVersion>
        <classifier>test-sources</classifier>
        <extension>jar</extension>
        <value>2.4.0-20180723.232410-177</value>
        <updated>20180723232411</updated>
      </snapshotVersion>
    </snapshotVersions>
  </versioning>
</metadata>

 
This behavior is very similar to this issue: https://issues.apache.org/jira/browse/MDEPLOY-221

Since 2.3.0 snapshots work with the same maven 3.3.9 version and maven deploy 2.8.2 plugin, it is highly possible that we introduced a new plugin that causes this. 

The most recent addition is the spot-bugs plugin, which is known to have incompatibilities with other plugins: https://github.com/spotbugs/spotbugs-maven-plugin/issues/21

We may want to try building without it to sanity check.

Issue Links

• relates to: SPARK-24529 Add spotbugs into maven build process
• links to: [Github] Pull Request #21865 (ericfchang)