I'm running Umbraco 4.5.2 with SQL Server Express on an Amazon EC2 small server instance, and perfoamance is typically excellent. Periodically, though, I see CPU utilization spiking to 100% on a W3WP process, and I can't trace it to one event in particular. I've looked through the umbraco system logs, the server's event viewer, and the IIS logs and there's doesn't seem to be an obvious culprit.
I've set up web.config as recommended for production mode (no version check, no trace, no debug).
Generally intermittent problems are tricky to trace, but any suggestions how to go about it?
IMHO the problem with working on virtualised environments is that the performance stats are unreliable - it could quite easily be the case that something else happening on another instance on your server is causing this spike. I'd copy the installation locally or on to a piece of dedicated hardware that you own and throw some load at it to see if you can reproduce the problem.
I have the installation running locally, and I never see this problem. When the production server's working right, it's very fast, but then it slows to a crawl and actions time out all over the place. This has lead to my umbraco.config file getting wiped out numerous time, giving the 'no nodes' error on my nice live site and necessitating a lengthy republish. It seemed to work great in development, but now that it's live I'm seeing this problem. The CPU load is 0-2% when it's working "normally".
Just my opinion but it sounds like a classic "Virtual environment" problem.
I'd love to help but there are way too many variables. I think if you couold produce the issue on a physical box then I'd have some basis to start with.
I believe I've found the problem - like in this thread, I had about 400 rules in the urlrewriting.net config file. I removed these and use the IIS URL Rewrite Module 2.0 instead, and am no longer seeing the usage spikes.
Andrew, how often were you experiencing the usage spikes before and under what kind of load? I think I may be experiencing something similar with an EC2 instance...
During normal operation, maybe once or twice a day, but when I logged in to the Umbraco backend it would happen every few minutes, and always after clicking a link, especially on a content node to bring up the editContent frame.
I found it helpful to be connected to the server via RDC and looking at the 'Worker Processes' window under IIS Manager. I'd refresh that when the CPU meter was pegged and it would have some process entries using the urlrewriting.net namespace.
Periodic 100% usage spikes
Hello,
I'm running Umbraco 4.5.2 with SQL Server Express on an Amazon EC2 small server instance, and perfoamance is typically excellent. Periodically, though, I see CPU utilization spiking to 100% on a W3WP process, and I can't trace it to one event in particular. I've looked through the umbraco system logs, the server's event viewer, and the IIS logs and there's doesn't seem to be an obvious culprit.
I've set up web.config as recommended for production mode (no version check, no trace, no debug).
Generally intermittent problems are tricky to trace, but any suggestions how to go about it?
IMHO the problem with working on virtualised environments is that the performance stats are unreliable - it could quite easily be the case that something else happening on another instance on your server is causing this spike. I'd copy the installation locally or on to a piece of dedicated hardware that you own and throw some load at it to see if you can reproduce the problem.
I have the installation running locally, and I never see this problem. When the production server's working right, it's very fast, but then it slows to a crawl and actions time out all over the place. This has lead to my umbraco.config file getting wiped out numerous time, giving the 'no nodes' error on my nice live site and necessitating a lengthy republish. It seemed to work great in development, but now that it's live I'm seeing this problem. The CPU load is 0-2% when it's working "normally".
Just my opinion but it sounds like a classic "Virtual environment" problem.
I'd love to help but there are way too many variables. I think if you couold produce the issue on a physical box then I'd have some basis to start with.
I believe I've found the problem - like in this thread, I had about 400 rules in the urlrewriting.net config file. I removed these and use the IIS URL Rewrite Module 2.0 instead, and am no longer seeing the usage spikes.
Andrew, how often were you experiencing the usage spikes before and under what kind of load? I think I may be experiencing something similar with an EC2 instance...
During normal operation, maybe once or twice a day, but when I logged in to the Umbraco backend it would happen every few minutes, and always after clicking a link, especially on a content node to bring up the editContent frame.
I found it helpful to be connected to the server via RDC and looking at the 'Worker Processes' window under IIS Manager. I'd refresh that when the CPU meter was pegged and it would have some process entries using the urlrewriting.net namespace.
is working on a reply...