Maven CI Best Practices
Yesterday our Admin came to me and told me, that the Harddrive on our CI Server is nearly full. This is due to the fact that our Nexus Repository as well as our Hudson CI Server are installed on the same machine. Hudson is using it’s own locale repository and therefor all Snapshots are filling up the diskspace.
Hudson has its own local repository due to the fact, that we are following the best practices mentioned by Jason Van Zyl on this blog post.
What I did now, was to purge the local repository of Hudson. Unfortunately we are not following all Maven best practices and therefor one project (the master pom for all other projects) is then missing from the local repository. I hope that I can clean this up during the next couple of days (hey, we are in a pretty tough project right now, and therefor the time for this is not really there). This means, that we do have to build and "mvn install" this project (base-build) directly after the purge. Furthermore I am cleaning up the Backups we are doing using ThinBackup (a small Hudson Plugin). Therefor we are using the following script:
#!/bin/sh -e
# fetches all changes in all repositories and sync it to redmine
cd /var/cache/hudson/mvn-repository
rm -rf *
cd /var/lib/hudson/jobs/base-build/workspace
su hudson -c 'mvn clean install' > /dev/null
cd /var/cache/hudson/backup
find . -mtime +5 -exec rm {} \;
Furthermore we are now using a scheduled Task in nexus to purge the Nexus Repository from any Snapshots older then 5 days, but keeping at least the latest snapshot (see another blog post).
After some more cleaning of e.g. log files and more, the disk is now half full (or half empty, depending on your point of view). This was more then enough for me.
Hope this helps.
Search
confused thoughts from a confused mind