Scheduled stop/start SQL instances

[FuriousDBA]

Hi David,
we have an SQL instance hosted on VM with scheduled stop/start. Each day I need to dismiss Instance uptime alert, but Snapshot Age for most of the day is Critical because it misses some DBA-Dash collections schedule (e.g. Server/Database principals, VLFs and other) which are scheduled by default at midnight.

1. is it possible to add a time range on instance level in which we assume the server should be running? This way, we could avoid uptime alerts.
2. is it possible to trigger gathering all collections once SQL Instance starts after some period of being down?
   Thx,

[DavidWiseman]

Hi,

You can adjust the schedules either at the service level or the instance level to ensure that the collections run when the instance is online.

DBA Dash doesn't really have uptime monitoring as such, though the collection dates can highlight when an instance is offline. It also highlights if a particular collection is failing - or maybe not running because the instance was offline when it's scheduled. The best way to check if an instance if offline at the moment would be to check the collection date for the Instance collection. This might be something I'll improve in a future version.

If you have messaging enabled in the service config tool, you can trigger the collections to run on demand. If you click snapshot age column on the summary tab it will take you to the collection dates tab for the instance. An option will be available to trigger all collections with a Warning or a Critical status. This is manual rather than automatic though. There are a couple of issues with doing this automatically that would need to be considered.

• Better uptime monitoring might need to be addressed first
• You might not want certain collections to run outside their scheduled execution time.
• In the case of remote services - the service doesn't know when a collection last run. It could potentially be pushed from the repository side with the messaging feature though.
• The collection might be failing - a mechanism would need to be in place to prevent re-triggering it over and over.

Thanks,

David