This guide walks through the steps to bring up a demonstration CORD "POD", running in virtual machines on a single physical server (a.k.a. "CORD-in-a-Box"). The purpose of this demonstration POD is to enable those interested in understanding how CORD works to examine and interact with a running CORD environment. It is a good place for novice CORD users to start.
NOTE: This guide describes how to install a simplified version of a CORD POD on a single server using virtual machines. If you are looking for instructions on how to install a multi-node POD, you will find them in the Physical POD Guide. For more details about the actual build process, look there.
You will need a target server, which will run both a development environment in a Vagrant VM (used to deploy CORD) as well as CORD-in-a-Box itself.
Target server requirements:
ubuntu user)If you do not have a target server available that meets the above requirements, you can borrow one on CloudLab. Sign up for an account using your organization's email address and choose "Join Existing Project"; for "Project Name" enter cord-testdrive.
NOTE: CloudLab is supporting CORD as a courtesy. It is expected that you will not use CloudLab resources for purposes other than evaluating CORD. If, after a week or two, you wish to continue using CloudLab to experiment with or develop CORD, then you must apply for your own separate CloudLab project.
Once your account is approved, start an experiment using the OnePC-Ubuntu14.04.5 profile on the Wisconsin, Clemson, or Utah clusters. This will provide you with a temporary target server meeting the above requirements.
Refer to the CloudLab documentation for more information.
On the target server, download the script that installs CORD-in-a-Box and run it. The script's output is displayed and also saved to ~/cord/install.out:
curl -o ~/cord-in-a-box.sh https://raw.githubusercontent.com/opencord/cord/master/scripts/cord-in-a-box.sh bash ~/cord-in-a-box.sh -t
NOTE: If you are connecting to a remote target server, it is highly recommended that you run the cord-in-a-box.sh script in a tmux session, or use mosh to connect to the target rather than ssh. Without one of these, interrupted connectivity between your local machine and the remote server may cause the CiaB install to hang.
The script takes a long time (at least two hours) to run. Be patient! If it hasn't completely failed yet, then assume all is well!
The script builds the CORD-in-a-Box and runs a couple of tests to ensure that things are working as expected. Once it has finished running, you'll see a BUILD SUCCESSFUL message.
The file ~/cord/install.out contains the output of the build process, post-bootstrap phase.
There is an -b option to cord-in-a-box.sh that will checkout a specific changeset from a gerrit repo during the run. The syntax for this is <project path>:<changeset>/<revision>. It can be used multiple times - for example:
bash ~/cord-in-a-box.sh -b build/platform-install:1233/4 -b orchestration/service-profile:1234/2"
will check out the platform-install repo with changeset 1233, revision 4, and service-profile repo changeset 1234, revision 2.
You can find the project path used by the repo tool in the manifest/default.xml file.
The -f option to cord-in-a-box.sh can be used to configure an ONOS fabric for CORD-in-a-Box. The fabric consists of two leaf and two spine switches, each running a CPqD OpenFlow software switch controlled by ONOS. The build process automatically generates a configuration file for the fabric and pushes it to ONOS. THIS FEATURE IS EXPERIMENTAL AND STILL UNDER DEVELOPMENT.
CORD-in-a-Box creates a virtual CORD POD running inside Vagrant VMs, using libvirt as a backend.
As access to the libvirt socket depends on being in the libvirtd group, you may need to to logout and back in to have your shell session gain this group membership:
~$ groups xos-PG0 root ~$ vagrant status Call to virConnectOpen failed: Failed to connect socket to '/var/run/libvirt/libvirt-sock': Permission denied ~$ logout ~$ ssh node_name.cloudlab.us ~$ groups xos-PG0 root libvirtd
Once you have done this, you can inspect the status of the VM's by setting the VAGRANT_CWD environmental variable to the path to the cord-in-a-box Vagrantfile's parent directory, then run vagrant status:
~$ export VAGRANT_CWD=~/cord/build ~$ vagrant status Current machine states: corddev running (libvirt) prod running (libvirt) switch not created (libvirt) leaf-1 running (libvirt) leaf-2 running (libvirt) spine-1 running (libvirt) spine-2 not created (libvirt) testbox not created (libvirt) compute-node-1 running (libvirt) compute-node-2 not created (libvirt) compute-node-3 not created (libvirt) This environment represents multiple VMs. The VMs are all listed above with their current state. For more information about a specific VM, run `vagrant status NAME`.
The corddev VM is a development machine used by the cord-in-a-box.sh script to drive the installation. It downloads and builds Docker containers and publishes them to the virtual head node (see below). It then installs MaaS on the virtual head node (for bare-metal provisioning) and the ONOS, XOS, and OpenStack services in containers. This VM can be entered as follows:
$ ssh corddev
The CORD build environment is located in /cord/build inside this VM. It is possible to manually run individual steps in the build process here if you wish; see the Physical POD Guide for more information on how to run build steps.
The prod VM is the virtual head node of the POD. It runs the OpenStack, ONOS, and XOS services inside containers. It also simulates a subscriber devices using a container. To enter it, simply type:
$ ssh prod
Inside the VM, a number of services run in Docker and LXD containers.
vagrant@prod:~$ docker ps CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES 043ea433232c xosproject/xos-ui "python /opt/xos/mana" About an hour ago Up About an hour 8000/tcp, 0.0.0.0:8888->8888/tcp cordpod_xos_ui_1 40b6b05be96c xosproject/xos-synchronizer-exampleservice "bash -c 'sleep 120; " About an hour ago Up About an hour 8000/tcp cordpod_xos_synchronizer_exampleservice_1 cfd93633bfae xosproject/xos-synchronizer-vtr "bash -c 'sleep 120; " 2 hours ago Up 2 hours 8000/tcp cordpod_xos_synchronizer_vtr_1 d2d2a0799ca0 xosproject/xos-synchronizer-vsg "bash -c 'sleep 120; " 2 hours ago Up 2 hours 8000/tcp cordpod_xos_synchronizer_vsg_1 480b5e85e87d xosproject/xos-synchronizer-onos "bash -c 'sleep 120; " 2 hours ago Up 2 hours 8000/tcp cordpod_xos_synchronizer_onos_1 9686909333c3 xosproject/xos-synchronizer-fabric "bash -c 'sleep 120; " 2 hours ago Up 2 hours 8000/tcp cordpod_xos_synchronizer_fabric_1 de53b100ce20 xosproject/xos-synchronizer-openstack "bash -c 'sleep 120; " 2 hours ago Up 2 hours 8000/tcp cordpod_xos_synchronizer_openstack_1 8a250162424c xosproject/xos-synchronizer-vtn "bash -c 'sleep 120; " 2 hours ago Up 2 hours 8000/tcp cordpod_xos_synchronizer_vtn_1 f1bd21f98a9f xosproject/xos "python /opt/xos/mana" 2 hours ago Up 2 hours 0.0.0.0:81->81/tcp, 8000/tcp cordpodbs_xos_bootstrap_ui_1 e41ccc63e7dd xosproject/xos "bash -c 'cd /opt/xos" 2 hours ago Up 2 hours 8000/tcp cordpodbs_xos_synchronizer_onboarding_1 7fdeb35614e8 redis "docker-entrypoint.sh" 2 hours ago Up 2 hours 6379/tcp cordpodbs_xos_redis_1 84fa440023bf xosproject/xos-postgres "/usr/lib/postgresql/" 2 hours ago Up 2 hours 5432/tcp cordpodbs_xos_db_1 ef0dd85badf3 onosproject/onos:latest "./bin/onos-service" 2 hours ago Up 2 hours 0.0.0.0:6653->6653/tcp, 0.0.0.0:8101->8101/tcp, 0.0.0.0:8181->8181/tcp, 0.0.0.0:9876->9876/tcp onosfabric_xos-onos_1 e2348ddee189 xos/onos "./bin/onos-service" 2 hours ago Up 2 hours 0.0.0.0:6654->6653/tcp, 0.0.0.0:8102->8101/tcp, 0.0.0.0:8182->8181/tcp, 0.0.0.0:9877->9876/tcp onoscord_xos-onos_1 f487db716d8c docker-registry:5000/mavenrepo:candidate "nginx -g 'daemon off" 3 hours ago Up 3 hours 443/tcp, 0.0.0.0:8080->80/tcp mavenrepo 0a24bcc3640a docker-registry:5000/cord-maas-automation:candidate "/go/bin/cord-maas-au" 3 hours ago Up 3 hours automation c5448fb834ac docker-registry:5000/cord-maas-switchq:candidate "/go/bin/switchq" 3 hours ago Up 3 hours 0.0.0.0:4244->4244/tcp switchq 7690414fec4b docker-registry:5000/cord-provisioner:candidate "/go/bin/cord-provisi" 3 hours ago Up 3 hours 0.0.0.0:4243->4243/tcp provisioner 833752cd8c71 docker-registry:5000/config-generator:candidate "/go/bin/config-gener" 3 hours ago Up 3 hours 1337/tcp, 0.0.0.0:4245->4245/tcp generator 300df95eb6bd docker-registry:5000/consul:candidate "docker-entrypoint.sh" 3 hours ago Up 3 hours storage e0a68af23e9c docker-registry:5000/cord-ip-allocator:candidate "/go/bin/cord-ip-allo" 3 hours ago Up 3 hours 0.0.0.0:4242->4242/tcp allocator 240a8b3e5af5 docker-registry:5000/cord-dhcp-harvester:candidate "/go/bin/harvester" 3 hours ago Up 3 hours 0.0.0.0:8954->8954/tcp harvester 9444c39ffe10 registry:2.4.0 "/bin/registry serve " 3 hours ago Up 3 hours 0.0.0.0:5000->5000/tcp registry 13d2f04e3b9b registry:2.4.0 "/bin/registry serve " 3 hours ago Up 3 hours 0.0.0.0:5001->5000/tcp registry-mirror
The above shows Docker containers launched by XOS (image names starting with xosproject). Containers starting with onos are running ONOS. There is also a Docker image registry, a Maven repository containing the CORD ONOS apps, and a number of microservices used in bare-metal provisioning.
vagrant@prod:~$ sudo lxc list +-------------------------+---------+------------------------------+------+------------+-----------+ | NAME | STATE | IPV4 | IPV6 | TYPE | SNAPSHOTS | +-------------------------+---------+------------------------------+------+------------+-----------+ | ceilometer-1 | RUNNING | 10.1.0.4 (eth0) | | PERSISTENT | 0 | +-------------------------+---------+------------------------------+------+------------+-----------+ | glance-1 | RUNNING | 10.1.0.5 (eth0) | | PERSISTENT | 0 | +-------------------------+---------+------------------------------+------+------------+-----------+ | juju-1 | RUNNING | 10.1.0.3 (eth0) | | PERSISTENT | 0 | +-------------------------+---------+------------------------------+------+------------+-----------+ | keystone-1 | RUNNING | 10.1.0.6 (eth0) | | PERSISTENT | 0 | +-------------------------+---------+------------------------------+------+------------+-----------+ | mongodb-1 | RUNNING | 10.1.0.13 (eth0) | | PERSISTENT | 0 | +-------------------------+---------+------------------------------+------+------------+-----------+ | nagios-1 | RUNNING | 10.1.0.8 (eth0) | | PERSISTENT | 0 | +-------------------------+---------+------------------------------+------+------------+-----------+ | neutron-api-1 | RUNNING | 10.1.0.9 (eth0) | | PERSISTENT | 0 | +-------------------------+---------+------------------------------+------+------------+-----------+ | nova-cloud-controller-1 | RUNNING | 10.1.0.10 (eth0) | | PERSISTENT | 0 | +-------------------------+---------+------------------------------+------+------------+-----------+ | openstack-dashboard-1 | RUNNING | 10.1.0.11 (eth0) | | PERSISTENT | 0 | +-------------------------+---------+------------------------------+------+------------+-----------+ | percona-cluster-1 | RUNNING | 10.1.0.7 (eth0) | | PERSISTENT | 0 | +-------------------------+---------+------------------------------+------+------------+-----------+ | rabbitmq-server-1 | RUNNING | 10.1.0.12 (eth0) | | PERSISTENT | 0 | +-------------------------+---------+------------------------------+------+------------+-----------+ | testclient | RUNNING | 192.168.0.244 (eth0.222.111) | | PERSISTENT | 0 | +-------------------------+---------+------------------------------+------+------------+-----------+
The LXD containers ending with names ending with -1 are running OpenStack-related services. These containers can be entered as follows:
$ ssh ubuntu@<container-name>
The testclient container runs the simulated subscriber device used for running simple end-to-end connectivity tests. Its only connectivity is to the vSG, but it can be entered using:
$ sudo lxc exec testclient bash
The compute_node-1 VM is the virtual compute node controlled by OpenStack. This VM can be entered from the prod VM. Run cord prov list to get the node name (assigned by MaaS). The node name will be something like bony-alley.cord.lab; in this case, to login you'd run:
$ ssh ubuntu@bony-alley
Virtual machines created via XOS/OpenStack will be instantiated on this compute node. To login to an OpenStack VM, first get the management IP address (172.27.0.x):
vagrant@prod:~$ source /opt/cord_profile/admin-openrc.sh vagrant@prod:~$ nova list --all-tenants +--------------------------------------+-------------------------+--------+------------+-------------+---------------------------------------------------+ | ID | Name | Status | Task State | Power State | Networks | +--------------------------------------+-------------------------+--------+------------+-------------+---------------------------------------------------+ | 3ba837a0-81ff-47b5-8f03-020175eed6b3 | mysite_exampleservice-2 | ACTIVE | - | Running | management=172.27.0.3; public=10.6.1.194 | | 549ffc1e-c454-4ef8-9df7-b02ab692eb36 | mysite_vsg-1 | ACTIVE | - | Running | management=172.27.0.2; mysite_vsg-access=10.0.2.2 | +--------------------------------------+-------------------------+--------+------------+-------------+---------------------------------------------------+
The VM hosting the vSG is called mysite_vsg-1 and we see it has a management IP of 172.27.0.2. Then run ssh-agent and add the default key (used to access the OpenStack VMs):
vagrant@prod:~$ ssh-agent bash vagrant@prod:~$ ssh-add
SSH to the compute node with the -A option and then to the VM using the management IP obtained above. So if the compute node name is bony-alley and the management IP is 172.27.0.2:
vagrant@prod:~$ ssh -A ubuntu@bony-alley ubuntu@bony-alley:~$ ssh ubuntu@172.27.0.2 # Now you're inside the mysite-vsg-1 VM ubuntu@mysite-vsg-1:~$
These VMs run software switches for the CORD fabric. In the default configuration they run standard Linux bridges. If you have chosen to run cord-in-a-box.sh with the experimental -f option, the VMs run CPqD switches controlled by ONOS running in the onosfabric_xos-onos_1 container.
You can access the MaaS (Metal-as-a-Service) GUI by pointing your browser to the URL http://<target-server>:8080/MAAS/. E.g., if you are running on CloudLab, your <target-server> is the hostname of your CloudLab node. The username is cord and the auto-generated password is found in ~/cord/build/maas/passwords/maas_user.txt. For more information on MaaS, see the MaaS documentation.
You can access the XOS GUI by pointing your browser to URL http://<target-server>:8080/xos/. The username is xosadmin@opencord.org and the auto-generated password is found in ~/cord/build/platform-install/credentials/xosadmin@opencord.org.
The state of the system is that all CORD services have been onboarded to XOS. You can see them in the GUI by clicking Services at left. Clicking on the name of a service will show more details about it.
A sample CORD subscriber has also been created. A nice way to drill down into the configuration is to click Customize at left, add the Diagnostic dashboard, and then click Diagnostic at left. To see the details of the subscriber in this dashboard, click the green box next to Subscriber and select cordSubscriber-1. The dashboard will change to show information specific to that subscriber.
The Kibana web interface to the ElasticStack log aggregation system can be found at: http://<target-server>:8080/kibana/.
On initial login, you will be asked to create an index for the logstash-* files - do this and then access the main logging interface under Discover. More information on using Kibana can be be found in its documentation.
After CORD-in-a-Box was set up, a couple of basic health tests were executed on the platform. The results of these tests can be found near the end of ~/install.out.
This tests the E2E connectivity of the POD by performing the following steps:
ping in the client to a public IP address in the InternetSuccess means that traffic is flowing between the subscriber household and the Internet via the vSG. If it succeeded, you should see some lines like these in the output:
TASK [test-vsg : Output from ping test] ****************************************
Thursday 27 October 2016 15:29:17 +0000 (0:00:03.144) 0:19:21.336 ******
ok: [10.100.198.201] => {
"pingtest.stdout_lines": [
"PING 8.8.8.8 (8.8.8.8) 56(84) bytes of data.",
"64 bytes from 8.8.8.8: icmp_seq=1 ttl=47 time=29.7 ms",
"64 bytes from 8.8.8.8: icmp_seq=2 ttl=47 time=29.2 ms",
"64 bytes from 8.8.8.8: icmp_seq=3 ttl=47 time=29.1 ms",
"",
"--- 8.8.8.8 ping statistics ---",
"3 packets transmitted, 3 received, 0% packet loss, time 2003ms",
"rtt min/avg/max/mdev = 29.176/29.367/29.711/0.243 ms"
]
}
This test builds on test-vsg by loading the exampleservice described in the Tutorial on Assembling and On-Boarding Services. The purpose of the exampleservice is to demonstrate how new subscriber-facing services can be easily deployed to a CORD POD. This test performs the following steps:
curl from the subscriber test client, through the vSG, to the Apache server.Success means that the Apache server launched by the exampleservice tenant is fully configured and is reachable from the subscriber client via the vSG. If it succeeded, you should see some lines like these in the output:
TASK [test-exampleservice : Output from curl test] *****************************
Thursday 27 October 2016 15:34:40 +0000 (0:00:01.116) 0:24:44.732 ******
ok: [10.100.198.201] => {
"curltest.stdout_lines": [
"ExampleService",
" Service Message: \"hello\"",
" Tenant Message: \"world\""
]
}
CORD-in-a-Box is a useful environment for integration testing and debugging. A typical scenario is to find a problem, and then rebuild and redeploy some XOS containers (e.g., a service synchronizer) to verify a fix. A workflow for quickly rebuilding and redeploying the XOS containers from source is:
~/cord/orchestration/xos*corddev VM and cd /cord/build./gradlew :platform-install:buildImages./gradlew -PdeployConfig=config/cord_in_a_box.yml :platform-install:publish./gradlew -PdeployConfig=config/cord_in_a_box.yml :orchestration:xos:publishAdditionally, if you made any changes to a profile (e.g., you added a new service), you'll need to re-sync the configuration from the build node to the head node. To do this run:
./gradlew -PdeployConfig=config/cord_in_a_box.yml PIprepPlatformNow the new XOS images should be published to the registry on prod. To bring them up, login to the prod VM and define these aliases:
CORD_PROFILE=$( cat /opt/cord_profile/profile_name ) alias xos-pull="docker-compose -p $CORD_PROFILE -f /opt/cord_profile/docker-compose.yml pull" alias xos-up="docker-compose -p $CORD_PROFILE -f /opt/cord_profile/docker-compose.yml up -d" alias xos-teardown="pushd /opt/cord/build/platform-install; ansible-playbook -i inventory/head-localhost --extra-vars @/opt/cord/build/genconfig/config.yml teardown-playbook.yml; popd" alias compute-node-refresh="pushd /opt/cord/build/platform-install; ansible-playbook -i /etc/maas/ansible/pod-inventory --extra-vars=@/opt/cord/build/genconfig/config.yml compute-node-refresh-playbook.yml; popd"
To pull new images from the database and launch the containers, while retaining the existing XOS database, run:
$ xos-pull; xos-up
Alternatively, to remove the XOS database and reinitialize XOS from scratch, run:
$ xos-teardown; xos-pull; xos-launch; compute-node-refresh
If the CORD-in-a-Box build fails, you may try simply resuming the build at the place that failed. The easiest way is to do is to re-run the cord-in-a-box.sh script; this will start the build at the beginning and skip over the steps that have already been completed.
If that doesn't work, the next thing to try is running cord-in-a-box.sh -c (specify the -c flag). This causes the script to clean up the previous installation and start from scratch.
If running cord-in-a-box.sh -c repeatedly fails for you, please tell us about it on the CORD Slack channel!
If you got this far, you successfully built, deployed, and tested your first CORD POD.
You are now ready to bring up a multi-node POD with a real switching fabric and multiple physical compute nodes. The process for doing so is described in the Physical POD Guide.