July 23, 2002 at 12:13 pm
This is a SQL Server6.5 related issue.
We recently started queries into the database over the web and since then we have had two incidents where we see the above error. When this happens You cant get into the Enterprize manager, you cant shut the SQL Server services. The only way out is to bounce the whole NT server. I can only do a sp_who in ISQL mode and what I see is lots of processes stacked up one after another and some of them in 'spinloop' status. Lots of processes are from the UNIX server coming in over the web. most of them in sleeping status.
Does any one know whats going on. What is a spinloop status - and why do we get stuck with max processes when most of the processes are in sleeping status but SQL Server counts them as live process and we run out of processes.
Any help will be appreicated.
I know we are in process of upgrading but we have to live with this version for a while.
Regards,
Naushi Hussain
Regards,
Naushi Hussain
July 23, 2002 at 4:23 pm
A few thoughts:
v6.5 has conneciton limits. This is enforced regardless of the state of the connection. Architectural limit, so you have to live with it. You can write a job to run every minute to kill the oldest xx processes based on some values.
EM has larger overhead, so get used to isql and get proficient.
Not sure about spinloop. Don't have BOL available for this version, but I suspect some type of blocking. Search "spinloop" on Technet and you will get lots of hits.
Steve Jones
July 23, 2002 at 4:27 pm
You probably also want to make sure you leave a copy of ISQL or EM open - otherwise if/when you hit the limit, you wont be able to get a connection to do anything about it!
Andy
July 23, 2002 at 4:35 pm
See articles at
http://www.ntfaq.com/Articles/Index.cfm?ArticleID=14402
http://support.microsoft.com/default.aspx?scid=KB;EN-US;Q271835&
Can you get the commands been send by these processes?
Viewing 4 posts - 1 through 3 (of 3 total)
You must be logged in to reply to this topic. Login to reply