Comment by DrJ

neptune.io's seems to be providing the solution that someone didn't fix during the day whenever an alert fires.

out of disk space -> clean up your disks (e.g. logrotate)

process crash -> should be using a process supervisor

low cpu utilization -> autoscaling