Re: Multi install with existing MAAS starts all services except for “IP Pending” on Glance Simplestreams Image Sync

Wed Jul 29 15:39:39 UTC 2015

Hi Jeff, so it looks like you're running vivid on the host, is that right?
Your point that the sync charm is the only one deployed as vivid is a good
catch - while this should still work, it might be causing problems for
juju's local provider.
I will fix the deploy code for the sync charm to use the same series as the
other charms, but one thing you can try in the meantime is use --series
vivid to just deploy the whole thing as vivid.
This has been tested and should work.

-mike

On Wed, Jul 29, 2015 at 8:25 AM, Jeff McLamb <mclamb at gmail.com> wrote:

> Hey Adam, Mike -
>
> The environment is up, but the container that should contain
> glance-simplestreams is not up at all. Here is the lxc-ls output from
> the node that contains all of the infrastructure containers:
>
> ubuntu at cooperative-crowd:~$ sudo lxc-ls --fancy
> NAME                      STATE    IPV4          IPV6  AUTOSTART
> ----------------------------------------------------------------
> juju-machine-3-lxc-0      RUNNING  192.168.1.17  -     YES
> juju-machine-3-lxc-1      RUNNING  192.168.1.18  -     YES
> juju-machine-3-lxc-2      RUNNING  192.168.1.19  -     YES
> juju-machine-3-lxc-3      RUNNING  192.168.1.20  -     YES
> juju-machine-3-lxc-4      RUNNING  192.168.1.21  -     YES
> juju-machine-3-lxc-6      RUNNING  192.168.1.22  -     YES
> juju-machine-3-lxc-7      RUNNING  192.168.1.23  -     YES
> juju-trusty-lxc-template  STOPPED  -             -     NO
> juju-vivid-lxc-template   STOPPED  -             -     NO
> ubuntu at cooperative-crowd:~$
>
> As you can see, the lxc-5 machine that should be running is not.
>
> When I look at openstack-status all of the services have a green
> checkmark and are [started] except for Glance Simplestreams which is
> in the [pending] stage.
>
> In the place of an IP address it says IP Pending and "waiting for
> initial status".
>
> I was wondering if I was running out of IP addresses in that range on
> the MAAS DHCP server, but there are only 8 containers and there are
> plenty of addresses left in that range.
>
> Here is the output of juju status where it is revealed that:
>
> 3/lxc/5:
>         agent-state-info: 'failed to retrieve the template to clone:
> template container
>           "juju-vivid-lxc-template" did not stop'
>         instance-id: pending
>         series: vivid
>
> For some reason this is the only container that it tries to install
> using the vivid series. All other containers use trusty.
>
> Here is the full juju status output:
>
> http://paste.ubuntu.com/11960193/
>
> Thanks!
>
> Jeff
>
> On Wed, Jul 29, 2015 at 11:15 AM, Adam Stokes <adam.stokes at canonical.com>
> wrote:
> > Jeff,
> >
> > Is the environment still up? Can you `juju ssh
> glance-simplestreams-sync/0`
> > and grab the unit logs from `/var/log/juju/`.
> >
> > CCing Mike for further investigation into that service
> >
> > On Wed, Jul 29, 2015 at 10:25 AM, Jeff McLamb <mclamb at gmail.com> wrote:
> >>
> >> Hey Adam -
> >>
> >> Sure, here it is:
> >>
> >> http://paste.ubuntu.com/11959888/
> >>
> >> Thanks!
> >>
> >> Jeff
> >>
> >>
> >> On Wed, Jul 29, 2015 at 10:07 AM, Adam Stokes <
> adam.stokes at canonical.com>
> >> wrote:
> >> > Hey Jeff,
> >> >
> >> > Can you paste.ubuntu.com your ~/.cloud-install/commands.log?
> >>
> >> >
> >> > On Wed, Jul 29, 2015 at 1:38 AM, Jeff McLamb <mclamb at gmail.com>
> wrote:
> >> >>
> >> >> Hello -
> >> >>
> >> >> Due to Adam’s suggestion in a previous thread about Autopilot not
> >> >> working properly with a proxy running (e.g. masa-proxy) on the MAAS
> >> >> server, I killed the environment and started over with the Multi
> >> >> option.
> >> >>
> >> >> Interesting aside: you must openstack-install -k and
> openstack-install
> >> >> -u or otherwise remove ~/.cloud-install entirely because if you just
> >> >> do a -k and do not remove the landscape related files, it will still
> >> >> try to install an Autopilot-based setup even if you choose Multi.
> >> >>
> >> >> Anyway, the Multi install progressed well, bootstrapping juju with
> the
> >> >> VM node, followed by deploying services across 3 other physical nodes
> >> >> that are all big RAM and big core.
> >> >>
> >> >> One physical node was chosen as dedicated compute, one as dedicated
> >> >> Neutron, and the third was chosen for infrastructure, deploying what
> >> >> should be 8 containers.
> >> >>
> >> >> In the openstack-status view I see that all services have “started”
> >> >> except for the Glance - Simplestreams Image Sync service which is to
> >> >> be deployed as Container 5 on the infrastructure machine (Machine 3
> in
> >> >> my case). It is stuck in “IP Pending” state.
> >> >>
> >> >> I have rebooted the infrastructure node and all lxc containers came
> >> >> back up, except a lxc-ls —fancy does not show the existence of
> >> >> juju-machine-3-lxc-5. It does have STOPPED juju-trusty-lxc-template
> >> >> and juju-vivid-lxc-template. The vivid one was actually RUNNING
> before
> >> >> I rebooted the system, but now it is STOPPED.
> >> >>
> >> >> I can get to the horizon dashboard GUI but I cannot login with a “An
> >> >> error occurred authenticating. Please try again later.”
> >> >>
> >> >> Any ideas how to get that last service running so I can hopefully log
> >> >> into the horizon GUI for the first time?
> >> >>
> >> >> Thanks,
> >> >>
> >> >> Jeff
> >> >>
> >> >> --
> >> >> ubuntu-openstack-installer mailing list
> >> >> ubuntu-openstack-installer at lists.ubuntu.com
> >> >> Modify settings or unsubscribe at:
> >> >> https://lists.ubuntu.com/mailman/listinfo/ubuntu-openstack-installer
> >> >
> >> >
> >
> >
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.ubuntu.com/archives/ubuntu-openstack-installer/attachments/20150729/fdb79899/attachment.html>