Major improvements (data-dir, python, locations)

[etj]

• Fixes Make the geoserver container populate the datadir, at runtime #57
  • removes geoserver-data-dir image
  • makes geoserver image deal with the population of the data dir
• removes python dependencies
  • move j2 temlpate to envsubst template
  • rewrite tasks.py into a bah script
• file locations refactoring
• scripts relocated into /usr/local/bin in image
• JAVA_OPTS cleanup

Needs changes in docker compose in geonode and geonode-project, so let's release it as a different tag

[Copilot]

Pull request overview

Copilot reviewed 12 out of 15 changed files in this pull request and generated 2 comments.

Comments suppressed due to low confidence (3)

docker/geoserver/scripts/entrypoint.sh:33

• This glob moves both "$TMP_EXTRACT/data/" and "$TMP_EXTRACT/data/.". In bash, the second glob expands to "." and ".." as well, which can cause errors and can attempt to move the parent directory into GEOSERVER_DATA_DIR. This can lead to incomplete/failed datadir initialization on first boot.
  docker/geoserver/scripts/entrypoint.sh:21
• GEOSERVER_DATA_DIR is expanded unquoted. While unlikely, an unexpected value (whitespace, glob chars) could cause mkdir to create the wrong path(s). Quoting is safer and consistent with the rest of the script.
  docker/geoserver/scripts/set_geoserver_auth.sh:48
• The escape_regex() implementation uses a character class containing "@/-" which is interpreted as a range in many regex engines, making the escaping behavior broader than intended (and potentially incorrect) if this helper is used. If you keep the function, fix the character class so '-' is treated literally.

[etj]

Cleaning up JAVA_OPTS:

Old:

ENV JAVA_OPTS="-Djava.awt.headless=true -Dgwc.context.suffix=gwc -XX:+UnlockDiagnosticVMOptions -XX:+LogVMOutput -XX:LogFile=/var/log/jvm.log -XX:MaxPermSize=512m -XX:PermSize=256m -Xms2048m -Xmx2048m -XX:+UseConcMarkSweepGC -XX:ParallelGCThreads=4 -Dfile.encoding=UTF8 -Djavax.servlet.request.encoding=UTF-8 -Djavax.servlet.response.encoding=UTF-8 -Duser.timezone=GMT -Dorg.geotools.shapefile.datetime=false -DGS-SHAPEFILE-CHARSET=UTF-8 -DGEOSERVER_CSRF_DISABLED=true -DPRINT_BASE_URL=http://geoserver:8080/geoserver/pdf -Xbootclasspath/a:/usr/local/tomcat/webapps/geoserver/WEB-INF/lib/marlin-0.9.3.jar -Dsun.java2d.renderer=org.marlin.pisces.MarlinRenderingEngine"

New:

ENV JAVA_OPTS="-Djava.awt.headless=true -Dgwc.context.suffix=gwc -XX:+UnlockDiagnosticVMOptions -XX:+LogVMOutput -XX:LogFile=/var/log/jvm.log -Xms2048m -Xmx2048m XX:ParallelGCThreads=4 -Dfile.encoding=UTF8 -Djavax.servlet.request.encoding=UTF-8 -Djavax.servlet.response.encoding=UTF-8 -Duser.timezone=GMT -Dorg.geotools.shapefile.datetime=false -DGS-SHAPEFILE-CHARSET=UTF-8 -DGEOSERVER_CSRF_DISABLED=true -DPRINT_BASE_URL=http://geoserver:8080/geoserver/pdf

• -XX:PermSize=256m and -XX:MaxPermSize=512m
  The Problem: The Permanent Generation (PermGen) space was entirely removed back in Java 8 and replaced by Metaspace.
• -XX:+UseConcMarkSweepGC
  The Problem: The Concurrent Mark Sweep (CMS) Garbage Collector was fully removed in Java 14.
• -Xbootclasspath/a:.../marlin-0.9.3.jar -Dsun.java2d.renderer=org.marlin.pisces.MarlinRenderingEngine
  The Problem: In older Java versions, GeoServer explicitly included the external Marlin rendering engine jar to speed up vector tiling and shape generation. Since Java 9, the Marlin Renderer has been baked directly into the standard JDK runtime framework. Furthermore, -Xbootclasspath restrictions are extremely strict under Java 17's encapsulation architecture.

Let's make sure these changes are also reported in the sample env files in GeoNode and projects