All we need is an easy explanation of the problem, so here it is.
SOS_SCHEDULER_YIELD when using
signal_wait_time_ms not be exactly the same as
wait_time_ms for the
SOS_SCHEDULER_YIELD wait type?
For other wait types signal wait is when a worker thread is waiting in the Runnable Queue on the scheduler. Therefore I conclude that signal wait should be the same as total wait time (=
SOS_SCHEDULER_YIELD when looking at
So my problem is that the two columns mentioned should have the same value in this DMV for this particular wait type.
More specifically, I am looking at a case where
signal_wait_time_ms is larger than
wait_time_ms, and I would like an explanation for that difference.
The example I am looking at is from the book "Pro SQL Server 2019 wait Statistics" by Van de Laar. In a screenshot on page 120, signal wait is higher for
How to solve :
I know you bored from this bug, So we are here to help you! Take a deep breath and look at the explanation of your problem. We have many solutions to this problem, But we recommend you to use the first method because it is tested & true method that will 100% work for you.
From the documentation for
sys.dm_os_wait_stats (emphasis mine):
|Column name||Data type||Description|
|wait_time_ms||bigint||Total wait time for this wait type in milliseconds. This time is inclusive of signal_wait_time_ms.|
If you see signal_wait_time_ms higher than wait_time_ms, then per the current documentation, that is a bug. You should report it to Microsoft if it’s causing you problems so they can investigate where things got off track. It could be rounding errors in the underlying data that lead to a slight difference (you didn’t mention how far off the two values were).
You said this:
the example I am looking at is from the book "Pro SQL Server 2019 wait Statistics" by Van de Laar. In a screenshot on page 120 signal wait is higher for SOS_SCHEDULER_YIELD.
Since the example is from a screenshot in a book, it’s quite possible that editorial processes resulted in invalid information making it into the final published result. In that case, I wouldn’t worry about it too much. You could contact the author to see if it’s possible or likely that something happened like that (maybe they were splicing screenshots from multiple tests together for simplicity).
Note: Use and implement method 1 because this method fully tested our system.
Thank you 🙂