SQL MP ver 6.0.6648.0
last night one of my monitored sql 2005 servers had a serious failure, which i'm sure the business did not like but which gave me a nice opportunity to compare the alerts and notifications generated by the old MOM 2005 agent and the SCOM 2007 R2 agent, both installed on this box. MOM 2005 generated 18 emails, SCOM 2007 generated 8. i'm not too concerned about most of the discrepancies, as 2005 is pretty noisy, but one that stands out is that the one database was in an offline/suspect state for over an hour, and while MOM 2005 generated more than one of
Description: The database "[omitted]" in the instance "MSSQLSERVER" is in a critical state. The state is: Suspect.
scom 2007 did not mention it once. and forgetting about the email and alerting part of it, the Health Explorer for this database shows 0 state changes in its history. as such, an availability report on the database for the past 48 hours also shows 100% availability. there *is,* however, a gap in the performance montoring of the database that corresponds to this outage.
an hour after this particular episode ended, we took another database on this same sql server offline, which scom 2007 immediately noticed and notified on:
Description: Database [omitted_LITE] in SQL Server instance MSSQLSERVER on computer [omitted.local] is offline/recovery pending/suspect/emergency
i'm not sure how to troubleshoot the fact that the first database never changed states in scom, which is my main concern. i got enough other alerts/notifications last night to *tell* me that something was broken on the server, but i think scom ought to be able to keep an accurate reading on the database state.