Disaster recovery of JFrog Artifactory

What I learned today — 23 March 2018

Our Artifactory instance ran out disk space and crashed. Even though I resized the persistent volume on our cloud provider I couldn’t get Kubernetes to recognize the change to the PVC.

To get it back up in order to make a backup to restore in a fresh installation, I deleted everything in the logs directory. BAD IDEA. That was the end of trying to get that instance back.

Eventually I spun up a fresh instance of Artifactory with a new, larger persistent volume and mounted the old PVC in a temporary location. After skipping through the setup wizard I doubtfully used the system import function and pointed it to the backup directory on the old volume, which I never configured. Fortunately the system was able to import everything from the default backup! I couldn’t believe my luck!

How to prevent running into this again:

• Configure disk usage thresholds. This will make Artifactory reject requests that will make it exceed the threshold.
• Configure your local snapshots repositories to keep a limited number of versions of the same snapshot. The default is infinite.
• Configure proper backups