Lots of blocking from SQLNS

[bro...@frnk.com]

Hi, everyone. I'm embarassed to admit that I set up a SQLNS app some
time ago, and it's been so long since I worked with it that I'm at a
loss for where to start troubleshooting.

A SQLNS server that I work with is experiencing an inordinately high
volume of blocking. In particular, the procedure
NSAssignNextWorkItemToDistributor seems to be blocking itself quite a
bit, presumably meaning that it's being launched more frequently than
it completes.

Can some kind soul give me a couple of tips:

- Could this procedure be bogging down due to a poor maintenance scheme
in the NS config? If so, what should I look for?

- Alternatively, could I have misconfigured it to run too frequrntly?
Again, what should I be looking for?

Thanks much
-Brooks

[Joe Webb]

Hi Brooks -

It's good to have apps that run without problem for so long without
problems that you almost forget about them. :)

I'm assuming you're using SQLNS 2.0 SP1. Are you running Standard
Edition or Enterprise?

What's the QuantumDurantion for the Generator?

What about for the Distributor? (When the Distributor component is
idle, you can control how often it checks for work items to process by
adjusting the <QuantumDuration> element in the <Distributor> node of
the ADF. This *only* effects the Distributor when it's idle. As long
as it's processing work items, it'll keep checking before going into
an idle state.)

How many Delivery Channels do you have defined for the protocols
supported by the Notification Class? The more Delivery Channels, the
more possible work items that will be created. Each is considered
independent and will be processed in parallel.

How many Distributors do you have defined? And what's the
ThreadPoolSize?

Have you adjusted the Notification Batch Size (assuming Enterprise
version)?

Other general performance considerations: Scheduled the vacuumer to
run periodically during non-peak times to clean up old data and
curtail distributor logging.

HTH...

--
Joe Webb
SQL Server MVP

~~~
Get up to speed quickly with SQLNS
http://www.amazon.com/exec/obidos/tg/detail/-/0972688811

I support PASS, the Professional Association for SQL Server.
(www.sqlpass.org)

[Shyam Pather [MSFT]]

Brooks,
Do you perhaps have extra delivery channels defined in your application that
aren't used (i.e. there are no/very few subscriber devices configured for
those delivery channels)? There is a bug in Notification Services SP1 where,
if you have defined many delivery channels in your ICF (config file) that
aren't used, the distributor mistakenly tries to grab new work items too
often. THis leads to the blocking you're describing.

We do have a hotfix available for this problem. You will need to contact
Product Support to obtain it.

Thanks
-shyam

--
Learn more about SQL-NS:
http://www.amazon.com/exec/obidos/tg/detail/-/0672326647/
---------------------------------------------
This posting is provided "AS IS" with no warranties, and confers no rights.
Use of included script samples are subject to the terms specified at
http://www.microsoft.com/info/cpyright.htm.
---------------------------------------------
<bro...@frnk.com> wrote in message
news:1115246625.1...@z14g2000cwz.googlegroups.com...

[Ram]

below two sps are participating in the deadlock
• NSAssignNextWorkItemToDistributor (Exclusive lock)
• NSMarkWorkItemDistributionStart (Update lock)
on the below indexes
indid name
------ -----------------------------
1 NSDistributorWorkItemsPK
10 status_distributor_workitems
on NSDistributorWorkItems table. this occurestwice or thrice in an hour. can anyone suggest the soln.

From http://www.google.co.in/search?hl=en&q=SQL+server+Notification+service:+NSAssignNextWorkItemToDistributor&meta=

Posted via DevelopmentNow.com Groups
http://www.developmentnow.com/g/