We've been seeing this behavior since a recent reinstall of 4.5.10 where, after a day or two, the Machine Agent stops reporting metrics even thought availability looks ok. Investigation found errors in the logs complaining about having too many open files when trying to run extensions/scripts. Looking at open file descriptors for the process, it appears that it is leaking file handles for its own logs. I'm curious if anyone else has seen similar behavior recently.
From a test machine:
root@mm03:/proc/23667/fd# ps -ef | grep appd
root 11569 806 0 22:30 pts/0 00:00:00 grep --color=auto appd
root 23667 1 7 Nov18 ? 03:46:54 /opt/appdynamics/machine-agent//jre/bin/java -Dlog4j.configuration=file:/opt/appdynamics/machine-agent/conf/logging/log4j.xml -jar /opt/appdynamics/machine-agent/machineagent.jar
root@mm03:/proc/23667/fd# ls -l /proc/23667/fd | wc -l
4097
root@mm03:/proc/23667/fd# ls -l /proc/23667/fd | grep "machine-agent.log" | wc -l
3922
