Hi @Koffi ,
yarn.nodemanager.log-dirs is where User Jobs Containers write the stdout, stderr and syslogs. Yarn has log aggregation which copies the log files from NodeManager local directory to HDFS and remove them.
In yarn-site.xml look for below property in yarn configs in ambari (the below is from my property)
And make sure dedicated disk is allocated. And one more chance of disks getting full is if more parallel containers running on a NodeManager and writing too much of intermediate data.
The most common cause of log-dirs are bad is due to available disk space on the node exceeding yarn's max-disk-utilization-per-disk-percentage default value of 90.0%.
Either clean up the disk that the unhealthy node is running on, or increase the threshold in yarn-site.xml.
if you are more interested, please read : https://blog.cloudera.com/resource-localization-in-yarn-deep-dive/