[ubuntu-cloud] Post access

[Gerardo Fontes]

Thanks Scott, I'll check it and post back next week with a proper topic.

Kind regards,

-- 
Gerard

On Fri, Mar 18, 2011 at 9:33 AM, Scott Moser <smoser at ubuntu.com> wrote:

> On Thu, 17 Mar 2011, Gerardo Fontes wrote:
>
> > Hi Scott,
> >
> > My apologies, I meant Lucid instead of Hardy(I was discussing something
> > about hardy at the time and my mind messed up) so a Lucid HVM is exactly
> > what I'm looking forward to, it's kind of sad to know HVM are restricted,
> I
> > will ask EC2 support to see if I can be able to create HVM images, in the
> > meantime I'm really happy and grateful that you will be releasing a daily
> > Lucid HVM, really, thank you! if there's something I can help with please
> > let me know.
>
> Well, there is an image available now, but it doesn't "just work".
>
> The image is
>   us-east-1 ami-3e08fa57 hvm/ubuntu-lucid-daily-amd64-server-20110318
>
> I launched the instance, but it doesn't come all the way up.
> I believe its blocked on cloud-init running , which is blocked waiting on
> eth0 to come up.  I think that this is either a need to get the right
> modules into the ramdisk (by modifying /etc/initramfs/modules), or it is
> also possible our -server kernel simply doesn't have the right
> modules/function built.
>
> So, I would greatly appreciate your poking at that and seeing if you can't
> get somethign working.  The process for debugging is, unfortunately,
> expensive as you basically have to start an instance, stop it, detach
> volume to another instance and try to fix it, re-attach volume, start
> instance... and each start costs $1.60.
>
> I'm probably not going to poke at this  myself in the near term, but that
> hvm instance will be there if you want to try to get it going or diagnose
> further.
>
> In the end, if you find its modules we need to get into the ramdisk (or
> otherwise loaded on early boot -- /etc/modules) we can probably just
> modify the build scripts to do that for hvm instances of lucid.
>
> If its a kernel config issue, we'll need to get a bug opened and SRU'd.
>
> Scott
>
>
> >
> > Sincerely,
> >
> > --
> > Gerard
> >
> > On Thu, Mar 17, 2011 at 3:30 PM, Scott Moser <smoser at ubuntu.com> wrote:
> >
> > > On Thu, 17 Mar 2011, Gerardo Fontes wrote:
> > >
> > > > Hi,
> > > >
> > > > I would like post access to this list and plan to collaborate for an
> ec2
> > > > cluster compute instance for hardy.
> > >
> > > Hi,
> > >   I had agreed with Ahmed to turn on the cluster compute instance
> builds
> > > for lucid as a sniff test.  If the un-tested build doesn't fail there
> > > might be a daily Lucid HVM instance type tomorrow sometime.
> > >   Hardy, though, is much older, and many, many
> > > things weren't so well worked out in our build and release process on
> EC2.
> > > It would be more difficult to make a Hardy cluster compute instance.
> > >   So I would really like to avoid doing so and supporting hardy on
> > > cluster compute.
> > >
> > > As I understand it, Amazon limits creation of "new" HVM images to
> > > privileged accounts (of which 'Canonical' has one), so you can't do the
> > > same process as we do.
> > >
> > >   That said, with the presence of a lucid cluster compute instance you
> > > may be able to basically do a downgrade to hardy, or potentially a
> switch
> > > of the root volume on an existing instance.
> > >   There is more infomatoin about HVM instances at
> > >
> > >
> http://docs.amazonwebservices.com/AWSEC2/latest/UserGuide/using_cluster_computing.html
> > >
> > >   You'll have to be creative, but I do think that a hardy HVM instance
> > > *could* be created, but you might end up wanting to use a newer kernel.
> > >
> > >   Scott
> > >
> >
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.ubuntu.com/archives/ubuntu-cloud/attachments/20110318/e0e29f9b/attachment.html>