What is normal for SELECT * FROM DEADLOCK_GRAPH errors in a day RRS feed

  • Question

  • Im not even remotely a DB admin. neither is my DB admin. is a deadlock normal? and if so how many deadlocks should one have in 10 minutes? an hour? etc. running 2016 server in VMware environment that complicates things im sure but just lots of dead locks shitty performance and im trying to get a understanding on what is normal for deadlocks primarily.
    Tuesday, September 24, 2019 3:53 PM


All replies

  • Deadlocks are caused by applications, not SQL Server.  This is not something you fix or even monitor at the SQL Server level.  

    It is up to the application getting the deadlocks to decide what to do with the error.  A deadlock graph can help the application developer determine why a deadlock is happening, but not how to fix it.

    Tuesday, September 24, 2019 6:33 PM
  • well there ya go. thats how little I know... from what i gather the dead lock is caused when the same data is accessed at the same time. I guess i was just curious is 5 deadlocks from an application in 5 minutes normal or is that a day or month ? 
    Tuesday, September 24, 2019 6:39 PM
  • There isn't a number to give for deadlocks which are acceptable from a SQL Server side.  They are a normal part of a multi-user system.

    The standard for an application, is to simply retry if it gets a deadlock.  So the existence of a deadlock may not even be of concern if the application retried and succeeded.

    Tuesday, September 24, 2019 6:48 PM
  • I suggest you read:

    The only solution for handling deadlocks is to find the problem in your code and then modify your processing to avoid deadlock situations. ...

    Tuesday, September 24, 2019 6:53 PM