On Fri, Jan 8, 2016 at 4:42 PM, Alexander Wels <awels@redhat.com> wrote:
On Friday, January 08, 2016 04:29:47 PM Nir Soffer wrote:
On Fri, Jan 8, 2016 at 1:13 PM, Nathanaël Blanchet <blanchet@abes.fr> wrote:
Hi all,
Since I upgraded engine to 3.6, I noticed that the webadmin takes a lot of ressources whatever is the browser. It can become very slow even for small actions, like changing tabs or editing a vm. The browser activity becomes intensive (100% of cpu) and the processor very hot with a increased fan activity. I suppose javascript to be responsible of this behaviour. Is there a way to reduce the resource allocated to the webadmin? (This is not a weakness of my laptop which is an i7 cpu with 16GB of RAM)
Sounds like https://bugzilla.redhat.com/1264809
It was closed because we could not reproduce it with current versions.
I would try first to upgrade to latest 3.6 version, and if you can reproduce it, please reopen this bug.
Alexander: can you advice on how to debug this?
Nir
If you are willing to spend some time helping us debug that would be great as obviously we are unable to reproduce. If you have Chrome, can you open the development tools (ctrl-shift-i) and go to the 'Profiles' tab. That should give you the option to 'Collect Javascript CPU Profile' among other things. Make sure you select that, and click start. Then go in the web admin and do something that makes your CPU spike. Let it go for a little bit doesn't have to be long.
Go back to the CPU profiles tab and click stop. That will generate a profile and that profile will include the percentage of CPU usage per method. It should be sorted by percentage. If you can take a screen shot of that and give me the exact rpm version, I can use that to de-obfuscate the method name and hopefully get some useful information.
You can also save such profile to file, and attach it to the bug. When I reorted this, this was also reproducible in Firefox, and both profiles are attached to the bug. Nir