2

Our file server has always had a performance blip at 2pm every day. Nothing major but annoying. Finally tracked down the cause, or potential cause, being a defrag running at 2pm every day. Event ID 258 starts to trickle in a few minutes after 2pm. It analyses each disk twice. The only schedule task I have configured for defrag is at 2300 every Friday. This is the only task listed under the scheduler. I can see other people online have had this issue but have had no solution as yet.

We are running Server 2008 R2 in a domain with the File Server role added. It is a VM and there is a backup agent from CA ArcServe running. This does not happen on any other server in our infrastructure that I can find and all of them are VMs and most have the same backup agent.

All in all I am baffled. Is there some secret task running or is there a way to blank the task scheduler completely so I can start afresh?

I have included pictures of the task scheduler areas that I am aware of.

EDIT: Well after a few months we have started to realise that this is probably now beginning to have a performance impact and not just when the analysis is run. According to the following blog post it seems likely that the high Metafile usage we are seeing and the memory warnings we are receiving are down to this 2pm defrag every day. Would still be very appreciative of any further suggestions on how to track this one down.

enter image description here

enter image description here

2
  • 1
    Look at "Task Scheduler Library" and see if someone has created a task in there to run a batch file or PowerShell script that could trigger a defrag
    – Travis
    Jun 13, 2013 at 12:30
  • been through the entire Task Scheduler Library and cannot see any powershell or batch files. I have the deleted our own scheduled task for the moment to see if that may be inadvertently causing it but will have to wait until tomorrow
    – user35213
    Jun 13, 2013 at 14:51

1 Answer 1

1

Well after much hit and miss troubleshooting finally tracked down the cause. It turns out that CA Arcserve backup agents were responsible for the analysis being kicked off despite there being no option in the agent to disable this behaviour. Am trying to find out from CA if this is expected and how to go about making it carry this out at a sensible hour and not in the middle of the bloody day! If in doubt, Process Monitor will turn up the information but just expect to spend a long time trawling through the filters!

You must log in to answer this question.