CPU usage high with no IP scanning enabled

[pcrockett]

Is it normal to have a 90-100% CPU usage on a dual core xeon server (4gb ram)even when just scanning one system at a time? It seems that whenever there are a few systems in the scanning queue, even if I set the server to scan one system at a time, the cpu usage for the lansweeperservice is 100%. If the queue is empty and I force scan one system, it only peaks at around 80% usage for a few moments but not sustained high usage. When the queue is full, it remains at 100% until the queue is empty which normally takes a few hours for only around 500 systems.

During that time, the web interface loads slowly initially in the morning but speeds up for all IT staff after the initial morning access attempt and runs fairly well even when the system is at 100% CPU usage.

Really the biggest problem with the slowness is that we run other network and maintenance utilities on that same server that run very slow during the morning hours. So we'd like to see the CPU usage reduced for that service somehow. Right now, there isn't much CPU usage difference between scanning 3 systems at a time and 100 systems at a time. It's just slow. The only difference between the first and the latter is the latter will zip through all the systems quite fast, wheras the 1-3 at a time will take a half a day. We've worked around it simply with patience thus far, but were wondering if there was a solution we could try or perhaps an issue with our implementation that causes it.

We are running everything over a gigabit network. Running on Server 2008R2, 64bit. We have no ip scans set up either.

Thanks!

[mvbenz]

where are the autorun and eventlog settings? I am not finding them and have the same issue.

[Hemoco]

mvbenz wrote:
where are the autorun and eventlog settings? I am not finding them and have the same issue.

In the configuration tool: waittime.

[Hemoco]

I'm guessing this is the problem: http://www.paulscomputerservice.net/articles/article.php?ID=98

[pcrockett]

Thanks for the input, I changed the autorun settings a couple of days ago and have been watching it. I'm now back to 20 active scans and it is using only around 2% of the CPU (with random spikes up to 60%) Disabling the autorun seems to have resolved the issue? I noticed disabling eventlog seemed to diminish the amount of bandwidth used as well. This one I have been playing with in the past as it did seem to lengthen the amount of time it took to scan a system significantly.

Not sure why on any of it, but something seems to have reduced the cpu load to a manageable amount again!

Thanks for everything!

[creinking]

pcrockett,
You might also try setting EVENTLOG to -1. I experienced the exact same issue as you and it was the downloading of everyone's event log that was killing the cpu.

[Hemoco]

Could you try the following:
- Set the waittime of autorun to -1
- restart the service

[pcrockett]

Service Version: 4.1.0.11

CPU usage is by the service not the SQL server.

I attached a couple screenshots within a few seconds of each other to show the fluctuations a little. These particular screenshots were taken this morning with 20 systems scanning at one time and 446 in queue. I then changed the number of systems to process at once to 1, restart the service, and the service stabilizes more around 60% usage and peaks at 80% occasionally.

What are the typical CPU usage numbers people experience for a 20 system at once scan?

[Hemoco]

pcrockett wrote:
Is it normal to have a 90-100% CPU usage on a dual core xeon server (4gb ram)even when just scanning one system at a time?

No.

What is your version number of lansweeperservice.exe?
Is the cpu time caused by lansweeperservice.exe or by your sql server?