Talking to DO support, it seems like the dropplets don't have any swap partition set up by default. Setting up some swap space on our droplet appeared to solve this problem for us.

Apparently it was running out of memory. A quick fix java -Xmx64m

Sounds right.

Also be mindful that even with constant utilization if a process doesn't have absolutely perfect garbage collection its RAM utilization will creep up over time, which is often why there will be a master daemon that launches children so that they could be routinely cleaned out.

I'm seeing similar problems - processes suddenly get killed with no apparent reason. This includes postgresql and rake spec (ruby on rails testing process)...

Reliving this thread,

I have the same situation. Sometimes the jetty process is simply stopped, and your log doesn't have any memory leak.

My droplet configuration is: 1GB RAM, 30GB disk...with Ubuntu 12.04 x64

My Jetty is 7.6, and this is your settings at /etc/default/jetty:

JAVA_HOME=/usr/lib/jvm/jdk1.7.0_45
JAVA=$JAVA_HOME/bin/java
JAVA_OPTIONS=" -server -Xms128m -Xmx512m -XX:+DisableExplicitGC -Dorg.eclipse.jetty.server.Request.maxFormKeys=900000 -Dorg.eclipse.jetty.server.Request.maxFormContentSize=500000"
JETTY_HOME=/usr/lib/jvm/jetty-distribution-7.6.13
JETTY_USER=root
JETTY_PORT=8080
JETTY_HOST=0.0.0.0
JETTY_LOGS=$JETTY_HOME/logs/

Any idea about this situation ? :/

Thanks in advance.

I believe that I've found the problem...just my droplet has 1GB of RAM, and jetty service was configured with 768mb form Xmx arg, the java process might have finished because cannot allocate all required memory.

The service is running +- 10 hours without interruption.