root@vcsa [ ~ ]# ps ax | grep vmon
[process ID] ? SLl [time] /usr/lib/vmware-vmon/vmon
root@vcsa [ ~ ]# ls -l /proc/[process ID]/fd | wc -l
1025
When vmware-analytics service is down, the num of vmon file discriptors will be increase gradually.
But in the witness node, vmware-analytics service don't up so the increase is not evitable.
Therefore, if the witness node is left running for a long period of time (500 days or more), the file descriptors will run out.
Workaround is reboot the witness node to release file descriptors.