rcbops / opencenter

http://www.rackspace.com/cloud/private/openstack_software/
Other
9 stars 12 forks source link

[DE67] Chef nodes created are not actual OC agent nodes depending on /etc/hosts #476

Closed kstev closed 11 years ago

kstev commented 11 years ago

Going through OpenCenter install. Everything seems to function normally. I'm able to move my controller node into "Infrastructure" and OpenCenter is happy. All nodes are green. I noticed that all actions were mysteriously fast and that my controller node did not seem to have any OpenStack packages installed. On the controller node, I run "chef-client" and notice chef was installed but the node was not assigned any roles:

root@srv1:~# chef-client Starting Chef Client, version 11.2.0 resolving cookbooks for run list: [] Synchronizing Cookbooks: Compiling Cookbooks... [2013-04-01T21:23:40+00:00] WARN: Node srv1 has an empty run list. Converging 0 resources Chef Client finished, 0 resources updated

So I check chef and find that I have 2 chef nodes created for each OpenCenter agent node: root@chef-server:~# /opt/chef-server/bin/knife node list srv1 srv1.blacksheep.com root@chef-server:~# /opt/chef-server/bin/knife node show srv1 Node Name: srv1 Environment: _default FQDN: IP: 192.168.200.101 Run List: Roles: Recipes: Platform: ubuntu 12.04 Tags: root@chef-server:~# /opt/chef-server/bin/knife node show srv1.blacksheep.com Node Name: srv1.blacksheep.com Environment: rpcs FQDN: IP: Run List: role[ha-controller1] Roles: Recipes: Platform: Tags:

I believe this is because of my mangled /etc/hosts : root@srv1:~# cat /etc/hosts 127.0.0.1 localhost.localdomain localhost srv1.blacksheep.com srv1

If I fix my /etc/hosts then this does not happen. root@srv1:~# cat /etc/hosts 127.0.0.1 localhost.localdomain localhost 192.168.200.101 srv1.blacksheep.com srv1

(remove chef nodes > remove open center node > remove chef and associated files > remove node opencenter files > re-run agent install on node > move node to "Infrastructure")

root@chef-server:~# /opt/chef-server/bin/knife node list srv1.blacksheep.com root@chef-server:~# /opt/chef-server/bin/knife node show srv1.blacksheep.com Node Name: srv1.blacksheep.com Environment: rpcs FQDN: srv1.blacksheep.com IP: 192.168.200.101 Run List: role[ha-controller1] Roles: ha-controller1, base, mysql-master, rabbitmq-server, keystone, keystone-api, glance-setup, glance-registry, glance-api, nova-setup, nova-network-controller, nova-scheduler, nova-api-ec2, nova-api-os-compute, cinder-setup, cinder-api, cinder-scheduler, nova-cert, nova-vncproxy, horizon-server, openstack-ha Recipes: osops-utils::packages, openssh, ntp, sosreport, rsyslog::default, hardware, osops-utils::default, mysql-openstack::server, erlang::default, rabbitmq-openstack::server, keystone::server, keystone::keystone-api, glance::setup, glance::registry, glance::api, nova::nova-setup, nova-network::nova-controller, nova::scheduler, nova::api-ec2, nova::api-os-compute, cinder::cinder-setup, cinder::cinder-api, cinder::cinder-scheduler, nova::nova-cert, nova::vncproxy, mysql::client, mysql::ruby, horizon::server, openstack-ha::default Platform: ubuntu 12.04 Tags:

My hostname remained the same throughout : root@srv1:~# hostname srv1.blacksheep.com root@srv1:~# cat /etc/hostname srv1.blacksheep.com root@srv1:~# hostname -s srv1

The only change was to /etc/hosts

claco commented 11 years ago

Actually, this isn't necessarily a hosts problem, but a chef-install problem with client.rb when the hostname changes. We should be specifying node_name in client.rb so it's not effected by hostname/hosts shenanigans.

See: https://github.com/rcbops/opencenter-agent/issues/229

claco commented 11 years ago

Fixed in https://github.com/rcbops/opencenter-agent/pull/242