Looking for a Maven Repository Cleanup script (unix) for Snapshot builds (diskspace)

So we have our own private Maven repository that we publish snapshot builds to.

We have lots builds so diskspace is starting to become a problem for all our snapshot builds. While its fun and all to go manually do this I was wondering if anyone knows of a CRON script that I can run to do the snap开发者_StackOverflow中文版shot cleanup.

I know sonatype does this for their own repo but I could not find a script.

To find all snapshot files that were updated more than two weeks ago:

 find . -type f -mtime +14 | grep SNAPSHOT

Pipe that to xargs rm and you should be good.

The one caveat: a repository manager will create a metadata.xml file that lists all published revisions. Assuming that you're just using scp to publish, and a webserver to retrieve, I don't think that file exists (so the fact that this script doesn't touch it shouldn't be an issue).

The following script works fine for me:

#!/bin/sh
REPO=/var/www/maven2/snapshots
find $REPO -type d -name '*-SNAPSHOT' | while read project; do
    if [ -f $project/maven-metadata.xml ]; then # Make sure this is a maven artifact directory
            # Assume that snapshot numbering is designed to be sorted numerically
            latestversion=$(ls $project | grep -v 'maven-metadata.*' | sort -n | grep '\.pom$' | tail -n1)
            latestversion=$(basename $latestversion .pom)
            # Delete everything, but the latest version and the maven metadata
            find $project -type f | grep -v -e 'maven-metadata.*' -e "$latestversion.*" | xargs rm
    fi
done