[Haizea] unmatching host CPU information between ONE and Haizea

[Shi Jin]

Thanks.
I am going to use subversion code then.

Before doing that, I did a quick test of changing the wakeup-intervals.
I did see 1 and 10 seconds working fine while 30 seconds does not come back.
Can you try 30 on your systems,  just in case?
Thanks a lot.

Shi



On Tue, Oct 19, 2010 at 12:25 PM, Borja Sotomayor <borja at borjanet.com>wrote:

> Hi,
>
> > First of all, the change you proposed does not work since q is a list.
> > I give it sum(q) instead
> > return "  |  ".join("%s: %i" % (res_type,sum(q)) for res_type, q in
> > self.quantity.items())
>
> ugh, sorry, I forgot to add that "%i" would have to be changed to "%s"
> (so the actual list is printed, that provides a better idea of the
> capacity of the individual instances).
>
> > It does print out 1600 instead of 100 in the haizea.log. But
> > haizea-list-hosts still shows 100.
>
> Uhm, ok, It looks like it's a problem with XML-RPC interface, not with
> the internal representation (i.e., Haizea is correctly storing
> multiple instances of 100, but the XML-RPC interface misrepresents
> this). I'll have to take a closer look at why only 1 vm gets deployed
> on each host (I'll need to see if I can reproduce the problem, which I
> can't do right now, but I'll try to get it done soon).
>
> > I have another question. If I set wakeup-interval:30 in the configuration
> > file, haizea does not wake up any more after the first
> > rounds. wakeup-interval:1 works as below. I am not sure whether this is a
> > related problem.
>
> That's odd. Did you do anything at the time Haizea stopped waking up?
> I usually have a wakeup interval of 10, and it seems to work fine for
> long-running processes.
>
>
> > By the way, I am using Haizea-1.0. Should I better use the subversion
> code
> > instead for later bug fix?
>
> The subversion code has a number of bug fixes in the core code and,
> unless you're using some of the new stuff (like deadline scheduling,
> or pricing policies, which is still in a rough state), the subversion
> code is preferable. I'd need to double-check the integration with
> OpenNebula, since I think some of the newer changes may have broken it
> but, if so, I'll be fixing them soon anyway so I can reproduce your
> problem.
>
> Cheers!
> --
> Borja Sotomayor
> -Scientific Writer at UChicago's Computation Institute
> http://people.cs.uchicago.edu/~borja/
> -Community Manager for the OpenNebula project
> http://www.opennebula.org/
> _______________________________________________
> Haizea mailing list
> Haizea at mailman.cs.uchicago.edu
> https://mailman.cs.uchicago.edu/mailman/listinfo/haizea
>



-- 
Shi Jin, Ph.D.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.cs.uchicago.edu/pipermail/haizea/attachments/20101019/75c5a561/attachment.htm