indexing tasks may remove already indexed content which differs only by case

Run Nexus 2.14.3 on Windows where the filesystem of the sonatype-work/nexus/storage is on a case-insensitive file system (NTFS).

Upload a Maven 2 artifact to a releases repo with this GAV:

g: example.app
a: fake
v: 1.0

Storage path on windows is

storage/releases/example/app/fake/1.0/fake-1.0.pom

maven-metadata.xml becomes:

<?xml version="1.0" encoding="UTF-8"?>
<metadata>
  <groupId>example.app</groupId>
  <artifactId>fake</artifactId>
  <versioning>
    <latest>1.0</latest>
    <release>1.0</release>
    <versions>
      <version>1.0</version>
    </versions>
    <lastUpdated>20170315142300</lastUpdated>
  </versioning>
</metadata>

Perform a Keyword Lucene search using term FAKE or fake. Both searches find the uploaded component.

Then upload

g: example.app
a: FAKE
v: 1.1

maven-metadata.xml becomes:

<?xml version="1.0" encoding="UTF-8"?>
<metadata>
  <groupId>example.app</groupId>
  <artifactId>FAKE</artifactId>
  <versioning>
    <latest>1.1</latest>
    <release>1.1</release>
    <versions>
      <version>1.0</version>
      <version>1.1</version>
    </versions>
    <lastUpdated>20170315142844</lastUpdated>
  </versioning>
</metadata>

Perform a Keyword Lucene search using term FAKE or fake. Both searches find the *2* results, both FAKE and fake.

A case-insensitive download is also possible. ie. This works:

http://localhost:8081/nexus/service/local/repositories/releases/content/example/app/fake/1.1/fake-1.1.pom

Now run update indexes task at the root of the releases repo.

This removes the already indexed GAVs that do not match actual paths stored on disk. ( ie uppercase FAKE ).

Now a keyword Lucene search using term FAKE or fake returns only *1* result for the `fake` artifact version 1.0.

Expected

• Update indexes or any indexing task should not remove already indexed artifacts that were not deleted from storage.
• Indexes should be consistent on Windows and Linux

Workaround

Users will not have this problem if the storage file system stores paths case sensitively. ( ie. Not Windows )