Check out Logstash or Splunk. Those can pipe your logs into an external database which can be used by a purpose-built UI for examining logs, and as a result it doesn't matter if the original files or machines are still around or not.
From: Lydia Ickler <ickle...@googlemail.com<mailto:ickle...@googlemail.com>> Date: Monday, December 19, 2016 at 7:38 AM To: <user@flink.apache.org<mailto:user@flink.apache.org>> Subject: Monitoring Flink on Yarn Hi all, I am using Flink 1.1.3 on Yarn and I wanted to ask how I can save the monitoring logs, e.g. for I/O or network, to HDFS or local FS? Since Yarn closes the Flink session after finishing the job I can't access the log via REST API. I am looking forward to your answer! Best regards, Lydia