Re: 答复: about xenalyze

[George Dunlap <george.dunlap@eu.citrix.com>]

On 05/07/2015 04:30 AM, 蒋雄伟(蒋冲) wrote:
>  Hi, George
> 
> I deploy a Discuz application in a VM and use jmeter to stress the application.  Meanwhile I use xentrace to trace.
> #xentrace -D -e 0x0008f000 -T 30 trace_file_discuz.bin
> 
> Then I use xenalyze to analyze
> #xenalyze --summary trace_file_discuz.bin It tells me as follows:
> 
>  --- Log volume summary ---
>   - cpu 0 -
>   hvm   :       1784
>   +-vmentry:        448
>   +-vmexit :        784
>   +-handler:        552
> 
> [...]
>  
>  
>  It seems different from what you show in your presentation 《Xenalyze Finding meaning in the chaos》。 What's wrong with it ?
>  I would like to see the reason, counts and cpu time for handler of each VMEXIT.  I would also like to see the wait time, blocked time , cpu usage of a domain.  What should I do?

The log volume summary is just the first part of the output; normally
the summary of stuff by domain is below that.

Can you please include the full output?

Thanks,
 -George


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel