Running the CDH 5.3.0 Quickstart VM; submitted a MR job and it was just sitting in YARN with no indication of progress.
Cloudera Manager shows lots of systems with log out of space errors so I looked through the filesystem for where I could free up space; found what appears to be excessive usage in the YARN usercache folder (6 GB in use).
File location is:
Can I just delete the cache manually or is there a better way? Is there a way to configure YARN to use less space?
Since you have already found the culprit I would recommend you to go go into CM and adjust the log retention settings for that service.
enable automatic yarn cache cleanup
yarn.nodemanager.localizer.cache.cleanup.interval-ms is 600000
also check your