July 7, 2023 at 11:26 pm
I just discovered trace flag 3444 set on a SQL 2017 srver, and I can't find reference to it. Has anyone come across this trace flag and what is it used for?
Thanks
MC
July 9, 2023 at 12:10 am
Thanks for posting your issue and hopefully someone will answer soon.
This is an automated bump to increase visibility of your question.
July 10, 2023 at 3:03 pm
I've come up with nothing in my searches, as well. Hopefully, someone will see this bit of a bump and know.
--Jeff Moden
Change is inevitable... Change for the better is not.
July 11, 2023 at 3:56 pm
I found out why the flag was set. This was a message from Microsoft.
Cause:
Your server encountered a non-yielding scheduler and this caused a stack dump on the primary node, which led to a lease timeout
Resolution:
The non-yielding scheduler was caused by a spinlock contention while waiting for a latch. Unfortunately, this is a limitation of SQL Server 2017 and though this issue is uncommon, and it can happen in certain situations. Should you experience this issue again, you can utilize Trace Flag 3444 as a workaround to increase the target recovery time on the database. If this is a large database, you could consider adding additional data files over different volumes as well. There has been a fundamental change in structure and processing in SQL Server 2019 with the increased features that allows more threads to prevent this type of contention, should you wish to consider upgrading to SQL Server 2019.
Viewing 4 posts - 1 through 3 (of 3 total)
You must be logged in to reply to this topic. Login to reply