One of our desktop support technicians noticed the following error from a 3rd party vendor application in an error log:

"9/22/2010 1:02 AM (GMT-06:00): The xxxxx Maintenance Service has reported an error. (200) CCmd: [DBNETLIB][ConnectionOpen (Connect()).]SQL Server does not exist or access denied.

Our technician e-mailed the error message (along with the fact that the database had recently been moved to a new server) to the vendor’s technical support team. Here is part of the response he got back from the vendor:

The 1:02am timing of that message makes me think that your backup is taking down the SQL Server while the xxxxx Maintenance Service is trying to hit it. This error message is pretty straightforward. SQL Server isn't there. And we know that it normally is.

You can either adjust your backup process or stop the xxxxx Maintenance Service on the xxxxx server temporarily while you run backups. If that's not the cause, we'll have to look into why your SQL Server is going down randomly.

Well, it turns out that the SQL Server error log on that instance shows that the instance has been running since August 17. The SQL Server Agent log shows that the nightly full native backup starts at 12AM, and completes in slightly over four minutes. That means that it is not even running when the vendor’s maintenance job is failing. Even if it were running, any SQL Server DBA should know that a SQL Server full backup does not cause blocking or locking, and it certainly does not “take down” SQL Server. As for SQL Server “going down randomly”, a quick peek at the SQL Error log knocks down that idea, since it shows the last time that the SQL Server Service was started.

You could also run this query (on SQL Server 2008 and above) to get the last time that SQL Server restarted:

-- Get last start time for SQL Server
SELECT sqlserver_start_time
FROM sys.dm_os_sys_info;

I let our technician know that the problem was probably a separate web.config that was still pointing at the old server location.