June 6, 2008 at 11:00 am
Every once in a while - no set pattern that I can find at all.. sqlservr.exe will jump up to 98-100% and stay there causing all the sites using that server to time out. The server throws no errors, and jobs even continue to run just very very slowly. We can go 3 weeks+ without incident.. sometimes a few months, but then it comes back... cpu peaks like its just working on something. Problem is I can't even get into activity monitor at that time to see whats running, deadlocks, etc... it times out attempting to open. A reset of the SQL service solves it every time and goes away until the next month.. or few weeks.
Ideas?
June 6, 2008 at 11:13 am
My first guess is that you've got a query or process that runs periodically, maybe something someone kicks off, that kills the server. Maybe a long cross join.
I'd run Profiler, just set it to rollover over 100MB or so, then whack the files if you have no issues. That way you can capture what was happening when it dies.
June 6, 2008 at 12:11 pm
Ok, I'll set up profiler that way... good advice. More or less that is what I was looking for - a way to be proactive so I could see what was happening when it does go belly up. We have pretty tight control over the environment, only 3 system admins have direct access to the box/sql installation. I know every job, and when it runs - and no one could have been preforming anything directly on the box.. it had to come through via... web site. One of the stored procs or queries in a web page, just not sure which one.
I'll let you know in a few weeks+ what I see.
Viewing 3 posts - 1 through 2 (of 2 total)
You must be logged in to reply to this topic. Login to reply