What could cause a mount to suddenly unmount?

I have been running glusterfs 3.4 in AWS for a couple months now, over the past month the glusterfs mount has disappeared from the gluster client server.

Stats from the client server seem normal,

The main gluster server will have a CPU spike at about 1am and then run normally for about 2 hours and then the mounts drops at 3am. This spike only goes to about 75% from the normal running CPU 50%

I do not see any change in the memory usage or disk i/o

This does not happen everyday however the 3 times it has happened 1am, 3am

I have checked crons on all of the servers, which there are none at 1 or 3.

I have checked all the logs on the main gluster server and cannot find anything to points me to what was happening on that server.

According to the cloudwatch logs it looks like the network traffic just drops off at 3am, which makes sense because there is no gluster traffic.

Any ideas?