Some times invariably the workflows go into loop due to mis-configuration by developers in these cases you can check the status of the workflows using
http://xxx:4502/etc/reports/workflowreport.html // This will report a very high number in the case above is true.
What can you to fix up the disk space in CQ in case above
Suggest you try using http://xxx:4502/system/console/jmx/com.adobe.granite.workflow%3Atype%3DMaintenance to purge completed worklows.
After that you might still need to run the repository garbage collection to free up the space (http://xxx:4502/system/console/jmx/com.adobe.granite%3Atype%3DRepository).
This is just a step to free up some space temporarily but ideally the configuration steps should be looked at again and fixed.