[Rdo-list] [openstack-packstack] Initial Kilo release

[Kaul, Yaniv]

Eventually made it working. For the benefit of all, here's how I've done it, include bizarre workarounds.


1. The Jenkins job is a shell script, essentially SSH'ing to the node and running a script.

SSH="ssh -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no -o GlobalKnownHostsFile=/dev/null"

${SSH} root@${CONTROLLER} "/home/public/scripts/devstack.sh"
echo "Running stack!"
${SSH} root@${CONTROLLER} "su - stack -c \"cd /opt/stack ; git config --global url.http://git.openstack.org/.insteadOf git://git.openstack.org/ ; ./stack.sh\""
${SSH} root@${CONTROLLER} "firewall-cmd --add-service http"


2. The /home/public/scripts/devstack.sh script:

firewall-cmd --add-service http || true

git config --global url.http://git.openstack.org/.insteadOf git://git.openstack.org/
cd /opt
git clone https://git.openstack.org/openstack-dev/devstack
/opt/devstack/tools/create-stack-user.sh
chown -R stack:stack /opt/devstack/
mv /opt/devstack /opt/stack

cat << 'EOF' >> local.conf
<your local.conf comes here>

EOF

sed -i 's/Defaults    requiretty/#Defaults    requiretty/' /etc/sudoers

mv local.conf stack
chown stack:stack stack/local.conf




Y.

> -----Original Message-----
> From: Kashyap Chamarthy [mailto:kchamart at redhat.com]
> Sent: Tuesday, February 24, 2015 11:12 AM
> To: Kaul, Yaniv
> Cc: Alan Pevec; Martin Mágr; Rdo-list at redhat.com
> Subject: Re: [Rdo-list] [openstack-packstack] Initial Kilo release
> 
> On Tue, Feb 24, 2015 at 03:52:18AM -0500, Kaul, Yaniv wrote:
> > > -----Original Message-----
> > > From: Kashyap Chamarthy [mailto:kchamart at redhat.com]
> > > Sent: Tuesday, February 24, 2015 10:23 AM
> > > To: Kaul, Yaniv
> > > Cc: Alan Pevec; Martin Mágr; Rdo-list at redhat.com
> > > Subject: Re: [Rdo-list] [openstack-packstack] Initial Kilo release
> > >
> > > On Mon, Feb 23, 2015 at 05:36:18PM -0500, Kaul, Yaniv wrote:
> > > > Anything is better than Devstack...
> > >
> > > Hmm, most (80%) of my test environment is via DevStack and I find it
> > > a huge time saver. Probably I just got used to it, I find it
> > > extremely quick (after the first
> > > run) to setup/tear-down environments -- just about 5 minutes or
> > > less. I know people running multi-node DevStack environments for
> > > rapid testing as well. :-)
> > >
> > > --
> > > /Kashyap
> >
> > By definition, pulling every single component from its latest greatest
> > upstream means it cannot be stable.
> 
> To avoid that, you can check out stable release of DevStack, which will inturn
> use only stable branches of the other OpenStack projects
> 
>     DevStack>$ git checkout remotes/origin/stable/juno
> 
> > In my specific case, it failed on heat - which I don't care about and
> > is not very useful to my work.
> 
> Also, to avoid issues like that the 'ENABLED_SERVICES' bit in local.conf is
> important.
> 
> You can just add the components that you use, it's been rock-solid for me that
> way. I use only components that I care about and absolutely nothing else --
> Nova, Glance, Neutron and Keystone.
> 
>     ENABLED_SERVICES=g-api,g-reg,key,n-api,n-cpu,n-sch,n-
> cond,mysql,rabbit,dstat,quantum,q-svc,q-agt,q-dhcp,q-l3,q-meta
> 
>     [NOTE: I also disable the 'n-cert' (Nova cert) service - which is
>     only for EC2, and is slated to be removed upstream.]
> 
> Since you care about Cinder too, so you can just add Cinder specific services in
> the ENABLED_SERVICES along with the above.
> 
> That's the localrc conf file I use:
> 
>     https://kashyapc.fedorapeople.org/virt/openstack/2-
> minimal_devstack_localrc.conf
> 
> Added benefit with the above config for me is also smaller footprint inside
> DevStack VM (with a single Nova instance running, I have about
> 1.3 GB of mem usage)
> 
>     https://kashyapc.fedorapeople.org/virt/openstack/heuristics/Memory-
> profiling-inside-DevStack.txt
> 
> You can compare what you see in your env by running the same $ ps_mem (as
> root). To install: $ yum install ps_mem
> 
> > I've tried disabling it (by adding 'disable_service heat h-api h-api-cfn h-api-cw
> h-eng' to my localrc) and then things broke even worse.
> > Re-trying...
> >
> > Here's my local.conf:
> > [[local|localrc]]
> > ADMIN_PASSWORD=123456
> > DATABASE_PASSWORD=$ADMIN_PASSWORD
> > RABBIT_PASSWORD=$ADMIN_PASSWORD
> > SERVICE_PASSWORD=$ADMIN_PASSWORD
> > SERVICE_TOKEN=a682f596-76f3-11e3-b3b2-e716f9080d54
> > #FIXED_RANGE=172.31.1.0/24
> > #FLOATING_RANGE=192.168.20.0/25
> > HOST_IP=10.103.233.161
> > CINDER_ENABLED_BACKENDS=xio_gold:xtremio_1
> >
> > [[post-config|$CINDER_CONF]]
> > [DEFAULT]
> > rpc_response_timeout=600
> > service_down_time=600
> > volume_name_template = CI-%s
> > enabled_backends=xtremio_1
> > default_volume_type=xtremio_1
> > [xtremio_1]
> > volume_driver=cinder.volume.drivers.emc.xtremio.XtremIOISCSIDriver
> > san_ip=vxms-xbrickdrm168
> > san_login=admin
> > san_password=admin
> > volume_backend_name = xtremio_1
> >
> >
> > Y.
> 
> --
> /kashyap