[Rt-devel] rt-4.0.0rc2 eating so much CPU
Evan_Horne at umit.maine.edu
Wed Jan 19 09:50:03 EST 2011
After upgrading from 3.9.7 to 4.0.0rc1 and then 4.0.0rc2, I was experiencing similar slowdown compared to 3.9.7.
I initially assumed the slow response was because I have RT currently running on a virtual machine on a desktop, but that doesn't explain why 3.9.7 was seemed so much more responsive on the same VM.
I grabbed the output with MasonX::Profiler enabled for both 4.0.0RC1 and 4.0.0RC2 while loading a page while this issue was occurring. I reverted my VM to an old state with RT3.9.7 installed and tried to enable MasonX::Profiler, but it wasn't
producing any extra output for some reason. I'd be happy to play around with it some more to get it working if anyone would find that information useful.
Devel mode is off in both of these instances. No more than a couple of tickets total in the database. A clean RT_SiteConfig file with nothing but MasonX::Profiler enabled doesn't seem to make any difference in page loads. The client is Firefox.
4.0.0 RC1 Output: http://pastebin.com/xUHP07nz
4.0.0 RC2 Output: http://pastebin.com/kFxr0Cg8
After playing around with RT 3.9.7 for a little bit and not getting the extra output from MasonX::Profiler, I reverted my VM back to it's current state with RT 4.0.0RC2 running and for some odd reason, it's now running faster than 3.9.7 was without
any additional changes. At this point I'd be willing to chalk up my own experience to a system configuration issue on my end, but if Odhiambo is still experiencing a similar issue, it may be related.
It never occurred to me to try Chrome's resource tracking feature while I was experiencing the issue to see where the delays were coming from on the client end until after it had resolved itself.
Just thought I'd throw out my experiences so far. If I can be of any further assistance, let me know.
Jesse Vincent <jesse at bestpractical.com> writes:
>Do you mean that you see that for the time that RT's homepage loads and then it drops back down or does it stick there?
>If the former, can you enable MasonX::Profiler (as described in RT_Config.pm) and see what it shows?
>Does it happen only for one page or for any page load?
More information about the rt-devel