I have an HDFS cluster where i get temporary Storage alerts that sometimes last hours.
When tracking the space consumption, i usually end up in the tmp/hive folder.
It looks like some job creates a very big amount of scratch data.
Is there any way to track the data to the hive job? Usually, if i go to the tez view, there are no jobs running, and it looks like most of the time, the temporary directory is not locked, for write.
I am running a HDP 2.6.4 cluster.