In our monitoring setup we know that some sensors goes in down state while the ops team is off-duty. These sensor might be down for 5 min, and we only notify on-call after 15 min.
I what to review all sensors that have transitioned to down state in an 12 hour window, to look for periodic errors.
We did have a nexus switch doing random stuff every 10-11 hour for a brief period. The ops team didn't catch the error, and only when the switch went into a boot loop we noticed :)