Re: Multi install with existing MAAS starts all services except for “IP Pending” on Glance Simplestreams Image Sync
Mike McCracken
mike.mccracken at canonical.com
Wed Jul 29 17:49:24 UTC 2015
Jeff, you can use --edit-placement to edit what goes where, and remove the
sync service.
Then you will just have to import a cloud image into glance yourself.
Or, you could let it try to deploy the sync service, then do a little bit
of tweaking and manual juju work to deploy the sync script as trusty
instead of vivid, on a default install where everything else is trusty.
on the host you're running the openstack installer from
export JUJU_HOME=~/.cloud-install/juju
since you have a failed sync service hanging around, first remove it with
this:
juju remove-service glance-simplestreams-sync
then look in ~/.cloud-install/local-charms/
there should be a directory 'vivid' with subdir glance-simplestreams-sync.
if you rename that directory to 'trusty', then re-deploy the sync charm
manually, it should work.
to redeploy, do this:
juju deploy --repository=~/.cloud-install/local-charms
local:trusty/glance-simplestreams-sync --to lxc:3 --config
~/.cloud-install/charmconf.yaml
-mike
On Wed, Jul 29, 2015 at 10:32 AM, Jeff McLamb <mclamb at gmail.com> wrote:
> Hey Mike -
>
> I am running MAAS Version 1.8.0+bzr4001-0ubuntu2 (vivid1) according to
> the version strip at the bottom of the MAAS Web GUI.
>
> Sure, will gladly file the bug report.
>
> At this point, in hopes of getting something up and running to a
> horizon dashboard I can log into, I am going to try to do the
> install-only mode of openstack-install, then manually select services
> but not glance sync. I presume I can get a working setup without sync,
> which I don't really need in the first place, right?
>
> Thanks,
>
> Jeff
>
>
>
> On Wed, Jul 29, 2015 at 1:26 PM, Mike McCracken
> <mike.mccracken at canonical.com> wrote:
> > Hi Jeff, your pastebin is showing that MAAS is complaining about the
> distro
> > series, and the installer is just forwarding that error along.
> > What version of MAAS are you running?
> >
> > As for the error with running a vivid container, that seems like a bug in
> > juju, and if you have the time, I'd appreciate it if you could file it
> as a
> > bug against lp:juju-core, including the info in your earlier email, in
> > particular the stuff around "Failed to mount cgroup at
> > /sys/fs/cgroup/systemd: Permission denied".
> >
> > It seems like there's a systemd-changeover related issue here that the
> juju
> > folks will want to know about.
> >
> > Thanks!
> > -mike
> >
> > On Wed, Jul 29, 2015 at 10:21 AM, Jeff McLamb <mclamb at gmail.com> wrote:
> >>
> >> This bug report seems to be related, if not the same thing:
> >>
> >> https://bugs.launchpad.net/landscape/+bug/1451385
> >>
> >> On Wed, Jul 29, 2015 at 1:10 PM, Jeff McLamb <mclamb at gmail.com> wrote:
> >> > Here is the output from trying with --series vivid:
> >> >
> >> > http://paste.ubuntu.com/11960777/
> >>
> >> >
> >> > I also did a re-install with --series trusty (really no different from
> >> > the first time) and everything turned out the same, except the order
> >> > in which containers came up and what got assigned to which container.
> >> >
> >> > However, logging into the physical node (machine 3) that hosts the
> >> > containers, I found the console log of the actual vivid template lxc
> >> > when it is in the RUNNING state and the glance sync container is the
> >> > only one left to provision.
> >> >
> >> > The errors below seem to make sense with respect to the complaint in
> >> > `juju status` that says it could not clone properly because the
> >> > template container did not terminate:
> >> >
> >> > ubuntu at cooperative-crowd:~$ sudo cat
> >> > /var/lib/juju/containers/juju-vivid-lxc-template/console.log
> >> > Failed to mount cgroup at /sys/fs/cgroup/systemd: Permission denied
> >> > systemd 219 running in system mode. (+PAM +AUDIT +SELINUX +IMA
> >> > +APPARMOR +SMACK +SYSVINIT +UTMP +LIBCRYPTSETUP +GCRYPT -GNUTLS +ACL
> >> > +XZ -LZ4 -SECCOMP +BLKID -ELFUTILS +KMOD -IDN)
> >> > Detected virtualization lxc.
> >> > Detected architecture x86-64.
> >> >
> >> > Welcome to Ubuntu 15.04!
> >> >
> >> > Set hostname to <juju-vivid-lxc-template>.
> >> > Initializing machine ID from random generator.
> >> > Failed to install release agent, ignoring: No such file or directory
> >> > Failed to create root cgroup hierarchy: No such file or directory
> >> > Failed to allocate manager object: No such file or directory
> >> > [!!!!!!] Failed to allocate manager object, freezing.
> >> >
> >> > On Wed, Jul 29, 2015 at 12:35 PM, Mike McCracken
> >> > <mike.mccracken at canonical.com> wrote:
> >> >> Can you share what goes wrong when you tried --series vivid ?
> >> >> It's always supported specifying whatever you like there, so it
> should
> >> >> work,
> >> >> and if something complained, that's a bug.
> >> >>
> >> >> Thanks,
> >> >> -mike
> >> >>
> >> >> On Wed, Jul 29, 2015 at 9:20 AM, Jeff McLamb <mclamb at gmail.com>
> wrote:
> >> >>>
> >> >>> It seems as though vivid is not a valid --series name in my version
> of
> >> >>> openstack-installer 0.99.19
> >> >>>
> >> >>> I can run it with trusty just fine, but I suspect that that will
> lead
> >> >>> to the same issue where glance is installed as vivid and the rest
> are
> >> >>> installed as trusty?
> >> >>>
> >> >>> Thanks,
> >> >>>
> >> >>> Jeff
> >> >>>
> >> >>>
> >> >>> On Wed, Jul 29, 2015 at 11:48 AM, Mike McCracken
> >> >>> <mike.mccracken at canonical.com> wrote:
> >> >>> > If you're able, I think the easiest way is to tear it down and
> >> >>> > re-deploy.
> >> >>> > Anything else would be a lot of manual steps. Juju has some
> upgrade
> >> >>> > capabilities, but the openstack installer pre-dates them, and so
> >> >>> > you'd
> >> >>> > have
> >> >>> > to hand-edit some config files, etc.
> >> >>> >
> >> >>> > -mike
> >> >>> >
> >> >>> > On Wed, Jul 29, 2015 at 8:44 AM, Jeff McLamb <mclamb at gmail.com>
> >> >>> > wrote:
> >> >>> >>
> >> >>> >> Cool, thanks for the tip, Mike.
> >> >>> >>
> >> >>> >> Yes the host is vivid running juju 1.24.3-vivid-amd64
> >> >>> >>
> >> >>> >> Given my current state, should I kill the environment and start
> >> >>> >> again,
> >> >>> >> or is there a way I can deploy or update what's currently there
> >> >>> >> without having to start over?
> >> >>> >>
> >> >>> >> Thanks,
> >> >>> >>
> >> >>> >> Jeff
> >> >>> >>
> >> >>> >>
> >> >>> >> On Wed, Jul 29, 2015 at 11:39 AM, Mike McCracken
> >> >>> >> <mike.mccracken at canonical.com> wrote:
> >> >>> >> > 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/20150729/7eb8d01c/attachment-0001.html>
More information about the ubuntu-openstack-installer
mailing list