I have been using the system_health extended event session to check for deadlocks, and thought this works well. However, recently, I turned on trace flags 1222 and 1204 to troubleshoot some deadlocking issues. What I found was that there were 13 more deadlocks caputred by the trace flag than the system_health EE session. And the most recent deadlock event was almost 3 days older than what was in the SQL Error log from the trace flags.
Is it possible that this EE session is missing deadlock events? Has anyone else seen this? Is this not a good resource for finding deadlock info?
Thanks
Matt