April 14, 2025 at 3:59 pm
I spotted this in the release notes for CU17 in SQL Server 2022:
KB3616559 - "Fixes a performance issue that you might encounter only when sp_lock is called frequently from multiple connections, which might cause a memory leak. The memory isn't cleaned up until you restart the SQL Server service.
Note: You need to turn on trace flag 15915."
We actually have been having an issue that feels like a memory issue since upgrading to SQL 2022, and have engaged MSFT and a few other "expert" SQL support engagements and nobody could figure it out. It's an instance that is acceptable to restart SQL on every weekend, so we've just been doing that, but I'm wondering if this is our issue.
I can't find any information from MSFT giving any information on this memory leak they are fixing, and if there is any way to tell by comparing anything before/after that trace flag being enabled to see if behavior changes. Has anyone happened to had this issue, and resolved it with this trace flag by chance?
The symptoms we see are essentially everything looking healthy, but every query (including a simple "select 1") just takes 5-10ms more than our normal baseline to execute across the whole instance. And a SQL service restart always fixes it. I think sp_lock is probably called by some of the DB monitoring tools we use, so not something we can easily just stop doing.
Thanks in advance for any info anyone may be able to share.
The Redneck DBA
April 15, 2025 at 4:10 pm
Thanks for posting your issue and hopefully someone will answer soon.
This is an automated bump to increase visibility of your question.
Viewing 2 posts - 1 through 1 (of 1 total)
You must be logged in to reply to this topic. Login to reply