Skip to content

using sm lite for xenial 16.04

sgurumurthy edited this page Jul 11, 2017 · 21 revisions

Build Details for current wiki

  • Server-Manager : R4.0.0.0-3060
  • Contrail-Cloud-Docker : R4.0.0.0-3057

Its fails in compute, due to mismatch of interface name/mac address, fix in 3060 of Contrail-Cloud-Docker

Pre-requisites:

  1. Node re-imaged with 16.04 (you may use lab-sm ubuntu-16.04-minimal image), 16.04.2/16.04.1 is not tested as of now. So avoid these.

    '''smlite works ONLY with ubuntu-16.04-minimal image'''

  2. Use lsb_release -a to verify installed version of OS

  3. Download 3060-sm-installer (See bottom of page for commands/url)

  4. Download 3057-xenial cloud images (4.0-3057-newton cloud images are last good contrail-cloud for xenial, see url at bottom of page)

  5. Get your testbed.py as sm-lite installation testbed.py

Patches/Known Issues:

  1. 3057-newton fails for interfaces names
  2. nova services doesn't comes up, needs to start manually
  3. testbed_parser.py may fail because of bug 1686245
  4. sm-webui installation fails for xenial, so --no-sm-webui
  5. re-provisioning of same nodes, would new cluster and new passwords every time, so it is advisable to use --cluster-id during provisioning,

Steps:

Steps needs to be run ALL nodes:

  1. apt-get install -y vlan : Install vlan package, first on ALL NODES ( This is temp step for 3060 release, else preconfig.py will fail)
  2. echo > /etc/apt/sources.list : clean-up your sources.list on ALL NODES, This step is temporary, and will moved to installation procedure.
  3. apt-get update : to ensure cleaned up repo list

Steps only for smlite node:

  1. Install server-manager installer packages ( dpkg -i contrail-server-manager-installer_4.0.0.0-3060~newton_all.deb ).
  2. Above step should create /opt/contrail/contrail_server_manager, go to /opt/contrail/contrail_server_manager
  3. Issue provision_containers.sh with testbed and contrail-cloud image. ./provision_containers.sh --testbed /root/testbed.py.dg-n002 --contrail-package /root/contrail-cloud-docker_4.0.0.0-3057-newton.tgz --no-sm-webui --cluster-id xenial-cluster (SM web-ui package has some trouble right now, disabling it)
  4. Above process would take a while to complete.
  5. Once above provision_containers.sh completed, it should print message about check status of provisioning.

----> Check provisioning status using /opt/contrail/contrail_server_manager/provision_status.sh

  1. You may check status using watch /opt/contrail/contrail_server_manager/provision_status.sh

Debugging/TroubleShooting:

Debugging is same as 14.04 SM-Lite, no changes.

  1. check for tailf /var/log/syslog | grep puppet-agent so check for puppet status. Once puppet completes, ansible provisioning starts.
  2. Openstack provisioning takes some time (may be ~15-20 mins, depends on hardware).
  3. Openstack provisioning may take 2 rounds of puppet run, wait 2nd run should be faster.
  4. Openstack provisioning should be completed with following message:

Apr 25 11:57:15 dg-n006 puppet-agent[14833]: (/Stage[openstack]/Contrail::Profile::Openstack_controller/Contrail::Lib::Report_status[openstack_completed]/Exec[contrail-status-openstack_completed]/returns) executed successfully

  1. Following means openstack provisioning has been completed, This is not a error.

Apr 25 11:58:07 dg-n006 puppet-agent[20636]: Could not retrieve catalog from remote server: Error 400 on SERVER: Failed when searching for node dg-n006.englab.juniper.net: Could not find a directory environment named 'TurningOffPuppetAgent__2017_04_25__11_57_43' anywhere in the path: /etc/puppet/environments. Does the directory exist?

  1. tailf /var/log/contrail-server-manager/debug.log provides details about ansible provisioning logs.
  2. tailf /var/log/contrail-server-manager/debug.log | grep "sm_ansible_callback.py" should filter ansible provisioning logs

"2017-04-25 12:06:05,901-INFO-sm_ansible_callback.py:43-append(): changed: [10.84.22.136]" "2017-04-25 12:06:05,908-INFO-sm_ansible_callback.py:43-append(): TASK [docker : Pin contrail apt repo "contrail" release to priority 999]" "2017-04-25 12:06:06,868-INFO-sm_ansible_callback.py:43-append(): changed: [10.84.22.136]"

Summary of Commands used:

wget http://10.84.5.120/github-build/mainline/3060/ubuntu-16-04/newton/artifacts/contrail-server-manager-installer_4.0.0.0-3060~newton_all.deb

wget http://10.84.5.120/github-build/mainline/3057/ubuntu-16-04/newton/artifacts/contrail-cloud-docker_4.0.0.0-3057-newton.tgz

apt-get install -y vlan

echo > /etc/apt/sources.list

apt-get update

dpkg -i contrail-server-manager-installer_4.0.0.0-3060~newton_all.deb

cd /opt/contrail/contrail_server_manager

./provision_containers.sh --testbed /root/testbed.py --contrail-package /root/contrail-cloud-docker_4.0.0.0-3057-newton.tgz --no-sm-webui --cluster-id xenial-cluster

watch /opt/contrail/contrail_server_manager/provision_status.sh

tailf /var/log/syslog | grep puppet-agent

tailf /var/log/contrail-server-manager/debug.log

Clone this wiki locally