SQL Server Performance Forum – Threads Archive

Excessive Handles for Agent Service

I seem to be having problems with the Agent Service running away with handles – the end result is that agent tasks stop working and the whole server slows down. I only have this issue on a couple of servers, from around 40. SQL2k ent sp3a w2k adv server – generally nodes on an active active cluster, but it has happened on a non clustered box.
Any pointers would be useful.
Thanks

What kind of error you get from log? Luis Martin
Moderator
SQL-Server-Performance.com The views expressed here are those of the author and no one else. There are no warranties as to the reliability or accuracy of anything presented here.

We don’t get any errors of any kind <grin > The problem shows as for instance:- right click on a job and run the job – nothing happens. Scheduled jobs just don’t run when they should.
We also noticed backups run through t sql commands from QA takes ages to run, very very slow.
Dropping into task manager shows lots of handles on the agent service, 3k has been lowest, 25k the highest so far. stopping and starting the service solves the problem.
There doesn’t seem to be any issues with memory at all. v strange

Try running Profiler to see what’s goin on with those jobs. Look for locks also.
Luis Martin
Moderator
SQL-Server-Performance.com The views expressed here are those of the author and no one else. There are no warranties as to the reliability or accuracy of anything presented here.

IS it the case that there are other jobs running on the system and they never finish at all. I would recommend looking at all the job steps and trying to exectte them using query analyzer. If that works fine, check if SQL Server Agent is started with some valid domain account. Gaurav
Moderator
Man thrives, oddly enough, only in the presence of a challenging environment- L. Ron Hubbard
The views expressed here are those of the author and no one else. There are no warranties as to the reliability or accuracy of anything presented here.

There are a large number of jobs on all the affected servers, I don’t actually see any common factor as our other cluster boxes are running essentially the same jobs without problems. ( I have 4 environments, test, stress, uat and production – my problem is with Stress and UAT )
We have a number of batch jobs that run constantly – these indicate they are running but actually are not when the problem arises, daily scheduled jobs just don’t run.
Agent service is a valid domain account ( different for each server ) I can’t seem to find any info about what should be a normal handle count, our prod cluster shows 280 handles whilst my two problem servers show over 8k and 3k. Don’t you just love these sorts of problems < grin >

Solved this – I do feel somewhat guilty but…… We’d rebuilt a node and the new nodes was at .818 and the existing at .760 it seems the handles is a byproduct of running an active-active with different patch version of sql 2k. ( Thanks to microsoft reading -UK )

This is the one of the reason to ensure all the machines involved must have similar levels of service pack and hotfixes. Satya SKJ
Moderator
http://www.SQL-Server-Performance.Com/forum
This posting is provided â€œAS ISâ€ with no rights for the sake of knowledge sharing.

]]>