Allow custom metrics to be optionally created as "event" alerts rather than default "continuous" alerts.
I have installed the "Used/available database space increased alert" which alerts when defined thresholds of used space have been exceeded. However, once activated, the alert remains in an active state with no additional notifications (easily missed). I'd prefer to have this metric run daily and notify each time a threshold is exceeded.
-
Thomas Franz commented
It would be nice to have a build in alert when someone (e.g. a vendor) created or dropped a database.
Reasons:
- you may want to delete old backup files (if dropped) or have to modify / delete some maintenances
- you'll make a manual full backup of the new DB so that log-backups could be taken by your backup solution
- maybe you have to add the new DB to monitoring (as DLM Dashboard), backups (and have to consinder its the disk space at the backup drive) or an Excel-Sheet with a list of your databases :-)Real-Live-Example:
I received at around 18:45 a log-backup-overdue-alert (should trigger if there were no log backups since 10 minutes). By checking the backup directory at the next morining I figured out that there were no log backups made between 16:00 and 00:45 (so the alert should have been triggered at 16:10). Speaking with the admin for this application I discovered that the DB was dropped at ~16:00 and recreated at ~18:35 (= database creation time in the log-backup-overdue-alert). Because they did not make a manual full backup it could not create log backups until the nightly job made an automatical full backup.If I had an alert telling me, that someone dropped / created a database, it would have been easier to figure out the "strange" behavior.
-
Chris (Tech Support) commented
Thank you for your suggestion. Yes, the alert will remain 'active' until the collected value drops below/above the specified lower threshold. (although you can configure it to send an email each time it crosses a threshold). If you clear the alert though, then the next time it will generate a new alert. But if the alert is unread or read but not cleared then the existing alert will be updated.
It was decided that this was best given that a custom metric which runs regularly (e.g. every minute) could conceivably spam a user with emails/alerts.
However, I have passed the suggestion to the development team to give the option of having 'event' behaviour. Once they have been able to review the idea, they may add it to their todo list.