- Downtime is set on a check and this downtime is visible in the UI
- Downtime exists for the check in the
- Notifications are sent for the check despite the above
Cause: disparity in downtime state between the database and datastore
When downtime is scheduled for a check or host, it is set in two places:
runtime.nagios_scheduleddowntime and the datastore of the collector(s) monitoring the host.
Downtime displayed in the UI comes from the
runtime.nagios_scheduleddowntime table. However, downtime stored in the collector datastore is what actually suppresses notifications.
If the datastore on the collector(s) monitoring the host is reinstalled or otherwise corrupted, the downtime is ‘forgotten’ by the collector so is not enforced.
Collector datastore corruption can happen when collectors are redeployed to different clusters.
recreate_downtime support script, downtime stored in the database can be recreated in the datastores of collectors.
See our page on using the