Differences

This shows you the differences between two versions of the page.

--- copper:academy:joid [2015/12/07 16:21]
Bryan Sullivan
+++ copper:academy:joid [2016/02/22 14:22] (current)
Bryan Sullivan
@@ Line 11: / Line 11: @@
   * Make sure you have good internet connectivity, as it's required for package install/updates etc.
-  * Install Ubuntu 14.04.03 LTS desktop on the jumphost. Use a non-root user (ubuntu, password ubuntu), which by default should be part of the sudoers group.
+  * Install Ubuntu 14.04.03 LTS desktop on the jumphost. Use a non-root user (opnfv, password you choose), which by default should be part of the sudoers group.
+    * If the jumphost is attached to a gateway with DHCP disabled (as described on the [[https://wiki.opnfv.org/copper/academy|Academy main page]], you will need to assign the static IP to the primary ethernet interface during Ubuntu install (since network autoconfig will fail). Use these values:
+      * IP address: 192.168.10.2
+      * Netmask: 255.255.255.0
+      * Gateway 192.168.10.1
+      * DNS: 8.8.8.8 8.8.4.4
+  * Install prerequisites
+<code>
+sudo apt-get update
+sudo apt-get upgrade -y
+sudo apt-get install bridge-utils vlan git openssh-server -y
+mkdir ~/.ssh
+chmod 744 ~/.ssh
+</code>
-==== If you don't have two interfaces on the jumphost ====
+  * Create a bridge brAdm on eth0, by modifying /etc/network/interfaces file on jumphost. Subnet "192.168.10.x" is used here to avoid conflicts with commonly used subnets in LAN environments.
-  * Note: this procedure may be buggy, and need more work to verify
-  * Set your wired connection to static IP. The examples below assume a configuration as used in /joid/ci/maas/att.
-    * Create a bridge brAdm on eth0, by modifying /etc/network/interfaces file on jumphost. Subnet "192.168.10.x" is used here to avoid conflicts with commonly used subnets in LAN environments.
 <code>
@@ Line 32: / Line 42: @@
 bridge_ports eth0
 </code>
-    * (maybe needed) Disable Ubuntu Network Manager
+    * If you installed Ubuntu desktop, you may need to disable Ubuntu Network Manager. In the case of static IP assignment during Ubuntu install as above, this will not be needed as the primary interface eth0 will be set to "unmanaged" by Network Manager.
 <code>
 sudo stop network-manager
@@ Line 38: / Line 48: @@
 </code>
     * Reboot
-==== Continuing with the normal install ====
-  * Install prerequisites
-<code>
-sudo apt-get update
-sudo apt-get upgrade -y
-sudo apt-get install bridge-utils vlan git openssh-server -y
-mkdir ~/.ssh
-chmod 744 ~/.ssh
-</code>
-  * Avoid a potential issue with incorrect locale setting (not sure if this was a mistake one time in Ubuntu install... will remove if skipping this step causes no issues normally)
-<code>
-export LANGUAGE=en_US.UTF-8
-export LANG=en_US.UTF-8
-export LC_ALL=en_US.UTF-8
-locale-gen en_US.UTF-8
-sudo dpkg-reconfigure locales
-</code>
   * Clone the joid repo
 <code>
+mkdir ~/git
+cd git
 git clone http://gerrit.opnfv.org/gerrit/joid.git
 </code>
-  * Set the correct MAC addresses on the controller and compute nodes in the template https://gerrit.opnfv.org/gerrit/gitweb?p=joid.git;a=blob;f=ci/maas/att/virpod1/deployment.yaml, under ~/joid/ci/maas/att/virpod1/deployment.yaml
+  * Set the correct MAC addresses on the controller and compute nodes in the template ~/joid/ci/maas/att/virpod1/deployment.yaml
 <code>
-vi ~/joid/ci/maas/att/virpod1/deployment.yaml
+vi ~/git/joid/ci/maas/att/virpod1/deployment.yaml
 </code>
   * Start MAAS and Juju bootstrap into two VMs on the jumphost
 <code>
-cd ~/joid/ci
+cd ~/git/joid/ci
-sudo ./02-maasdeploy.sh attvirpod1
+./02-maasdeploy.sh attvirpod1
-</code>
-  * First time, you may see an error as below. Logout and login again and run the same command. (Why this occurs will be described here...)
-<code>
-libvirt: XML-RPC error : Failed to connect socket to '/var/run/libvirt/libvirt-sock': Permission denied>
 </code>
@@ Line 87: / Line 73: @@
 </code>
-  * You should now be able to login to MAAS at http://192.168.10.3/MAAS (ubuntu/ubuntu). However there may have been issues in the commissioning of the jumphost - verify the status of the jumphost in the MAAS Nodes screen is "Ready". If not, the commissioning may have failed due to networking issues, e.g. brief internet service disruptions that can interfere with process steps. In that case, just restart the 02-maasdeploy.sh command.
+  * You should now be able to login to MAAS at http://192.168.10.3/MAAS (ubuntu/ubuntu). However there may have been issues in the commissioning of the jumphost - verify the status of the jumphost in the MAAS Nodes screen is "Ready". If you can't login to MAAS, or the jumphost node status is not "Ready", the commissioning may have failed due to networking issues, e.g. brief internet service disruptions that can interfere with process steps. In that case, just restart the 02-maasdeploy.sh command.
+    * Some debugging hints for the MAAS VM
-  * Install wakeonlan in maas node (workaround until this is merged into the juju-deployer tool)
+      * ssh ubuntu@192.168.10.3
-<code>
+      * tail -f /var/log/cloud-init-output.log
-ssh 192.168.10.3
-sudo apt-get install wakeonlan
-</code>
-  * As a temporary workaround for wakeonlan not being installed in the maas VM image, create the controller and compute machines manually thru the MAAS gui at http://192.168.10.3/MAAS (ubuntu/ubuntu)
+  * As a temporary workaround for current issues with node creation by MAAS, create the controller and compute machines manually thru the MAAS gui at http://192.168.10.3/MAAS (ubuntu/ubuntu)
-    * From the "Nodes" panel, select "Add Hardware/Machine"python-neutronclient, Machine name "controller1", MAC Address per your controller's MAC, Power type "Wake-On-LAN", MAC Address per your controller's MAC, and click "Save Machine".
+    * From the "Nodes" panel, select "Add Hardware/Machine", Machine name "node1-control", MAC Address per your controller's MAC, Power type "Wake-On-LAN", MAC Address per your controller's MAC, and click "Save Machine".
-      * You should see the machine power-on. You will be returned to the "Nodes" panel. Select "controller1.maas", "Edit", set Tags "control", and click "Save Changes".
+      * You should see the machine power-on. You will be returned to the "Nodes" panel. Select "node1-control", "Edit", set Tags "control", and click "Save Changes".
-    * Repeat for your compute node, naming it "compute1", specify its MAC, and set the Tags to "compute"
+    * Repeat for your compute node, naming it "node2-compute", specify its MAC, and set the Tags to "compute"
-  * When both nodes are in the "Ready" state in the MAAS gui (they will have been powered off by MAAS in after completion), enable IP configuration for controller1's second NIC (eth1). This is a workaround; setup of this NIC should be automated (how TBD)
+  * When both nodes are in the "Ready" state in the MAAS gui (they will have been powered off by MAAS in after completion), enable IP configuration for the controller's second NIC (eth1). This is a workaround; setup of this NIC should be automated (how TBD)
-    * From the "Nodes" panel, selection "controller1" and scrll down to the "Network" section. For "IP address" of eth1, select "Auto assign".
+    * From the "Nodes" panel, select "node1-control" and scroll down to the "Network" section. For "IP address" of eth1, select "Auto assign".
   * Run the OPNFV deploy via Juju
 <code>
-sudo ./deploy.sh -o liberty -s odl -t nonha -l attvirpod1
+cd ~/git/joid/ci
+./deploy.sh -o liberty -s odl -t nonha -l attvirpod1
 </code>
   * if the deploy times out and you output such as https://etherpad.opnfv.org/p/copper-juju-deploy-timeout
-    * debug, correct the issue and run
+    * debug, correct the issue, clean the juju environment as below, power-off the controller and compute nodes, and rerun the previous step
 <code>
-sudo ./clean.sh
+./clean.sh
-sudo ./deploy.sh -o liberty -s odl -t nonha -l attvirpod1
+# use the following commands if clean.sh hangs
+rm -rf ~/.juju/environments
+rm -rf ~/.juju/ssh
 </code>
@@ Line 127: / Line 113: @@
   * However, it's likely not yet complete as the juju charms take a while (30 minutes or more) to fully configure and bring up the services to the "active" status. You can see the current status from the command:
 <code>
-sudo juju status --format=tabular
+watch juju status --format=tabular
-[Services]
-NAME                  STATUS      EXPOSED CHARM
-ceilometer            waiting     false   local:trusty/ceilometer-44
-ceilometer-agent                  false   local:trusty/ceilometer-agent-20
-ceph                  active      false   cs:trusty/ceph-42
-cinder                maintenance false   cs:trusty/cinder-31
-cinder-ceph                       false   cs:trusty/cinder-ceph-14
-glance                waiting     false   local:trusty/glance-150
-heat                  waiting     false   local:trusty/heat-12
-juju-gui              unknown     false   cs:trusty/juju-gui-41
-keystone              waiting     false   local:trusty/keystone-0
-mongodb               unknown     false   cs:trusty/mongodb-28
-mysql                 unknown     false   cs:trusty/mysql-31
-neutron-api           waiting     false   local:trusty/neutron-api-1
-neutron-api-odl                   false   local:trusty/neutron-api-odl-0
-neutron-gateway       active      false   local:trusty/neutron-gateway-64
-nodes-api             active      false   local:trusty/ubuntu-nodes-controller-1
-nodes-compute         active      false   local:trusty/ubuntu-nodes-compute-1
-nova-cloud-controller maintenance false   local:trusty/nova-cloud-controller-501
-nova-compute          active      false   local:trusty/nova-compute-133
-ntp                               false   cs:trusty/ntp-14
-odl-controller        unknown     false   local:trusty/odl-controller-0
-openstack-dashboard   waiting     false   local:trusty/openstack-dashboard-32
-openvswitch-odl                   false   local:trusty/openvswitch-odl-0
-rabbitmq-server       active      false   local:trusty/rabbitmq-server-150
-(followed by more detail)
 </code>
-  * When it completes, you can get the addresses of the installed services by:
+  * Once all the services are either in state "active" or "unknown" (which just means that the JuJu charm for the service does not yet implement status reporting), install should be complete. When it completes, you can get the addresses of the installed services by:
 <code>
 juju status --format=short
@@ Line 186: / Line 145: @@
 </code>
-  * To see the detailed status:
-<code>
-juju status
-</code>
 ===== Installing Additional Tools =====
@@ Line 208: / Line 163: @@
 </code>
-===== Verifying Services are Operational =====
+===== Verifying OpenStack and ODL Services are Operational =====
 ) On the jumphost browse to, and verify that these services are active (e.g. per the example above):
@@ Line 224: / Line 179: @@
 </code>
-) Create image cirros-0.3.3-x86_64
+===== Verifying Services are Operational - Smoke Tests =====
+The following procedure will verify that the basic OPNFV services are operational.
+  * Clone the copper project repo:
+<code>
+mkdir ~/git
+cd ~/git
+git clone https://gerrit.opnfv.org/gerrit/copper
+</code>
+  * Execute "smoke01.sh", which uses the OpenStack CLI to
+    * Create a glance image (cirros)
+    * Create public/private network/subnet
+    * Create an external router, gateway, and interface
+    * Boot two cirros instances on the private network
+<code>
+cd ~/git/copper/tests/adhoc
+source smoke01.sh
+</code>
+  * From Horizon / Project / Compute / Instances, select "cirros-1" and "Console", and login with account "cirros" / "cubswin:)". Ping address of cirros-2 as shown in Horizon, 192.168.10.1 (external router), and opnfv.org (to validate DNS operation).
+  * Execute "smoke01-clean.sh" to delete all the changes from smoke01.sh.
+<code>
+cd ~/git/copper/tests/adhoc
+source smoke01-clean.sh
+</code>
+===== Verifying Services are Operational - Horizon UI and CLI =====
+This was the part-manual procedure developed before the smoke tests. This section will be maintained so that a Horizon procedure is also documented.
+) Create image cirros-0.3.3-x86_64
   * Horizon: From Project / Compute / Images, select "Create Image". Set options Name "cirros-0.3.3-x86_64", Image Location "http://download.cirros-cloud.net/0.3.3/cirros-0.3.3-x86_64-disk.img", Format "QCOW2", Architecture "x86_64", (leave other options as-is or blank), and select "Create Image".
   * CLI:
 <code>
-glance --os-image-api-version 1 image-create --name cirros-0.3.3-x86_64 --disk-format qcow2 --location http://download.cirros-cloud.net/0.3.3/cirros-0.3.3-x86_64 -disk.img --container-format bare
+glance --os-image-api-version 1 image-create --name cirros-0.3.3-x86_64 --disk-format qcow2 --location http://download.cirros-cloud.net/0.3.3/cirros-0.3.3-x86_64-disk.img --container-format bare
 +------------------+--------------------------------------+
 | Property         | Value                                |
@@ Line 252: / Line 240: @@
 </code>
 ) On jumphost, create external network and subnet using Neutron CLI
   * NOTE: Assumes you have completed steps in "Installing Additional Tools"
 <code>
@@ Line 288: / Line 276: @@
 | ipv6_address_mode |                                                    |
 | ipv6_ra_mode      |                                                    |
-| name              |                    neutron router-create                                |
+| name              |                                                    |
 | network_id        | b3913813-7a0e-4e20-b102-522f9f21914b               |
 | subnetpool_id     |                                                    |
@@ Line 295: / Line 283: @@
 </code>
 ) Create internal network.
   * Horizon: From Project / Network / Networks, select "Create Network". Set options Name "internal", select "Next", Subnet Name "internal", Network Address "10.0.0.0/24", select "Next" (leave other options as-is or blank), Allocation Pools "10.0.0.2,10.0.0.254", DNS Name Servers "8.8.8.8", select "Create" (leave other options as-is or blank).
   * CLI:
@@ Line 339: / Line 327: @@
 </code>
 ) Create router and external port
   * Horizon: From Project / Network / Routers, select "Create Router". Set options Name "external", Connected External Network: "public", and select "Create Router" (leave other options as-is or blank).
   * CLI:
@@ Line 376: / Line 364: @@
 </code>
 ) Add internal network interface to the router.
   * Horizon: From Project / Network / Routers, select router "external", and select "Add Interface". Select Subnet "internal", and select "Add Interface" (leave other options as-is or blank). Note: you may get disconnected from the Controller node for a short time
   * CLI:
@@ Line 383: / Line 371: @@
 </code>
 ) Launch images
   * Horizon: From Project / Compute / Images, select "Launch" for image "cirros-0.3.3-x86_64". Select Name "cirros-1", network "internal", and "Launch". Leave other options as-is or blank. Repeat for instance "cirros-2".
   * CLI:
@@ Line 459: / Line 447: @@
 </code>
 ) Verify connectivity from VMs.
   * Horizon: From Project / Compute / Instances, select "cirros-1" and "Console", and login with account "cirros" / "cubswin:)". Ping address of cirros-2 as shown in Horizon, 192.168.1.1 (external router), and opnfv.org (to validate DNS operation).

Wiki

User Tools

Site Tools

Differences

Page Tools