Re: [clif] exception on probe deployment

[Bruno Dillenseger <bruno.dillenseger@xxxxxxxxxxxxxxxxxxxx>]

olivier gusching wrote:

> Thanks for your help but i still got an exception -another one- which 
> occurs when trying to deploy a single probe (without injector) on a 
> server :
>
>     [java] NoSuchInterfaceException: Resource probe control0 in 
> component /null
>     [java]     at 
> org.objectweb.fractal.julia.generated.C9d834a99_0.bindFc(MIXIN[CompositeBindingControllerImpl]:6102) 
>
>     [java]     at 
> org.objectweb.fractal.julia.generated.Ce7005b70_0.bindFc(INTERFACE[BindingController]) 
>
>     [java]     at 
> org.objectweb.fractal.api.control.BindingController_Skel.send(Unknown 
> Source)
>     [java]     at 
> org.objectweb.fractal.rmi.RmiProtocol$ServerSession_Low.send(RmiProtocol.java:496) 
>
>     [java]     at 
> org.objectweb.jonathan.libs.protocols.tcpip.TcpIpProtocol$Session.run(TcpIpProtocol.java:453) 
>
>     [java]     at 
> org.objectweb.jonathan.libs.resources.JScheduler$JJob.run(JScheduler.java:244) 


I fixed this bug, but it seems that there are other bugs in this 
particular case of deploying probes alone without injector (it seems, at 
least, that some bindings are missing with the probe controller 
component). I may try to fix this, but I'm not sure. As a matter of 
fact, I'm working on an improved architecture in which all this probe 
specific code will be dropped, so...

> By the way, I have two remarks about the system monitoring :
>
> -Sometimes, the system monitoring stops during a bench (no measures 
> are taken any more)
> -The graph scale of the system monitoring is wrong -seems to be twice 
> as the perf monitoring-. Let's explain that : if you specify the 
> duration of the test into the display period, you will have a correct 
> graph for the performance monitoring, but the system monitoring graph 
> scale will not fit to the performance one, as the graph will exceed 
> the window width. 

All this monitoring part has to be refactored too. This will be done 
when we'll have new probes (e.g. from LeWYS or other contribution?).
Until we get such a new version, my advice is: use the combined 
performance+ressource monitoring window only, and forget about the two 
individual monitoring windows which are worse than the combined one.
If you really need to improve the combined window right now, you'll have 
to go into the source code, and probably propose a contribution... :o)

Best regards,
-- Bruno.