mthumphrey (8/14/2013)
I just figured out that I am having this same issue. the system_health EE session has fewer deadlock events that what is logged by trace flags 1204, 1222.Has anyone got more info on this?
I've opened a case with Microsoft, and I have gotten them to admit this is an issue.
They told me that they recommend me using the ERRORLOG info for troubleshooting deadlocks, but I am not letting them off the hook so easily. There is real value in being able to track this info programmatically and save it in a table for historic analysis, and doing that from the ERRORLOG just doesn't cut it. Besides, what is the use of the system_health session then, if it is not reliable? Might as well not have it running at all!
So I'm trying to get them to escalate this as a bug to their dev team.
Waiting to hear back...
Are you also seeing progressively newer deadlocks appearing on a day-by-day basis, but always a few days behind the current date?
It's as if the info is all in memory but the process writing to the system_health output is very slow and is not able to catch up.
Also, what version of SQL are you running?
__________________________________________________________________________________
SQL Server 2016 Columnstore Index Enhancements - System Views for Disk-Based Tables[/url]
Persisting SQL Server Index-Usage Statistics with MERGE[/url]
Turbocharge Your Database Maintenance With Service Broker: Part 2[/url]