It would be extremely helpful to have an alert that could hit on a delayed or hung database sync between the pollers and database. Perhaps it could reference 'Last Database Update', or be based on some other metric related to a poller not updating to the database. Being able to set an alert for a database update not happening for more than 60 minutes would be a great help.
This is a pretty critical health check of NPM and can be very elusive if not looking for it. Without some sort of alert, it can go unchecked and render the monitoring solution, or significant portions of it, pretty useless during the time of the issue.
There have been several instances over several years where administrators will discover that one of their pollers has ceased updating to the database. Unless you are looking for it or notice that you are missing statistics from some where, it is very difficult to tell that there is an issue. We have three pollers counting our NPM server and have experienced it twice now. The first time it took about a week before anyone noticed it. The second time it took about 12 hours because we just happened to be troubleshooting some CRC errors on an interface and noticed that our statistics had dropped off about 12 hours before. Two of the three pollers were good to go. Thwack has several instances of admins coming across a similar issue during upgrades.
Help in this alerting will be greatly appreciated! Thanks.
