On 09/08/11 - 01:37:18PM, Ian Main wrote:
On Tue, Sep 06, 2011 at 09:03:58AM -0400, Chris Lalancette wrote:
> On 09/02/11 - 05:12:48PM, Richard Su wrote:
> > Hi Chris,
> >
> > Just some quick feedback after playing with this a little bit. With
> > vSphere, the vm starts but state in conductor remains in "pending".
With
> > RHEV, the instance fails to start with this error in thin.log:
>
> Thanks for the test Richard, it is appreciated. Ian, can you take a look at
> this as part of your partial rework of the patch? Thanks.
Yeah, will do. That's great, thanks Richard.
Ian
I fixed this part, and posted another patch series. There are 2 new issues
that have cropped up:
1) jprovazn is getting some weird database errors. He suspects that it
might be related to sharing the same database connection in multiple
processes, and he very well might be right. What I'd like to avoid is loading
the rails environment *every* time we fork a process, so if you could look at
maybe just tearing down the database connection and reconnecting at the start
of the child process, that might fix this particular problem.
2) rwsu is running into an issue where RHEV-M doesn't work. This is because
we don't have the logic to monitor a RHEV-M instance that goes from
PENDING->STOPPED, and then starting it. I would probably suggest doing this
in dbomatic as well, but I'll leave it to you guys. rwsu is taking a look at
it now, so you should coordinate with him.
--
Chris Lalancette