26 KiB
Overview
In order to drive towards a production-ready Openstack solution, our goal is to provide containerized, yet stable persistent volumes that Kubernetes can use to schedule applications that require state, such as MariaDB (Galera). Although we assume that the project should provide a “batteries included” approach towards persistent storage, we want to allow operators to define their own solution as well. Examples of this work will be documented in another section, however evidence of this is found throughout the project. If you have any questions or comments, please create an issue.
IMPORTANT: Please see the latest published information about our application versions.
| Version | Notes | |
|---|---|---|
| Kubernetes | v1.5.1 | Custom Controller for RDB tools |
| Helm | v2.1.3 | Planning for v2.2.0 |
| Calico | v2.0 | calicoctl v1.0 |
| Docker | v1.12.1 | Per kubeadm Instructions |
Other versions and considerations (such as other CNI SDN providers), config map data, and value overrides will be included in other documentation as we explore these options further.
The installation procedures below, will take an administrator from a new kubeadm installation to Openstack-Helm deployment.
Kubernetes Preparation
This walkthrough will help you set up a bare metal environment with 5 nodes, using kubeadm on Ubuntu 16.04. The assumption is that you have a working kubeadm environment and that your environment is at a working state, prior to deploying a CNI-SDN. This deployment procedure is opinionated only to standardize the deployment process for users and developers, and to limit questions to a known working deployment. Instructions will expand as the project becomes more mature.
If you’re environment looks like this, you are ready to continue:
admin@kubenode01:~$ kubectl get pods -o wide --all-namespaces
NAMESPACE NAME READY STATUS RESTARTS AGE IP NODE
kube-system dummy-2088944543-lg0vc 1/1 Running 1 5m 192.168.3.21 kubenode01
kube-system etcd-kubenode01 1/1 Running 1 5m 192.168.3.21 kubenode01
kube-system kube-apiserver-kubenode01 1/1 Running 3 5m 192.168.3.21 kubenode01
kube-system kube-controller-manager-kubenode01 1/1 Running 0 5m 192.168.3.21 kubenode01
kube-system kube-discovery-1769846148-8g4d7 1/1 Running 1 5m 192.168.3.21 kubenode01
kube-system kube-dns-2924299975-xxtrg 0/4 ContainerCreating 0 5m <none> kubenode01
kube-system kube-proxy-7kxpr 1/1 Running 0 5m 192.168.3.22 kubenode02
kube-system kube-proxy-b4xz3 1/1 Running 0 5m 192.168.3.24 kubenode04
kube-system kube-proxy-b62rp 1/1 Running 0 5m 192.168.3.23 kubenode03
kube-system kube-proxy-s1fpw 1/1 Running 1 5m 192.168.3.21 kubenode01
kube-system kube-proxy-thc4v 1/1 Running 0 5m 192.168.3.25 kubenode05
kube-system kube-scheduler-kubenode01 1/1 Running 1 5m 192.168.3.21 kubenode01
admin@kubenode01:~$
Deploying a CNI-Enabled SDN (Calico)
After an initial kubeadmn deployment has been scheduled, it is time to deploy a CNI-enabled SDN. We have selected Calico, but have also confirmed that this works for Weave, and Romana. For Calico version v2.0, you can apply the provided Kubeadm Hosted Install manifest:
admin@kubenode01:~$ kubectl apply -f http://docs.projectcalico.org/v2.0/getting-started/kubernetes/installation/hosted/kubeadm/calico.yaml
PLEASE NOTE: If you are using a 192.168.0.0/16 CIDR for your Kubernetes hosts, you will need to modify line 42 for the cidr declaration within the ippool. This must be a /16 range or more, as the kube-controller will hand out /24 ranges to each node. We have included a sample comparison of the changes here and here.
After the container CNI-SDN is deployed, Calico has a tool you can use to verify your deployment. You can download this tool, calicoctl to execute the following command:
admin@kubenode01:~$ sudo calicoctl node status
Calico process is running.
IPv4 BGP status
+--------------+-------------------+-------+----------+-------------+
| PEER ADDRESS | PEER TYPE | STATE | SINCE | INFO |
+--------------+-------------------+-------+----------+-------------+
| 192.168.3.22 | node-to-node mesh | up | 16:34:03 | Established |
| 192.168.3.23 | node-to-node mesh | up | 16:33:59 | Established |
| 192.168.3.24 | node-to-node mesh | up | 16:34:00 | Established |
| 192.168.3.25 | node-to-node mesh | up | 16:33:59 | Established |
+--------------+-------------------+-------+----------+-------------+
IPv6 BGP status
No IPv6 peers found.
admin@kubenode01:~$
It is important to call out that the Self Hosted Calico manifest for v2.0 (above) supports nodetonode mesh, and nat-outgoing by default. This is a change from version 1.6.
Preparing Persistent Storage
Persistent storage is improving. Please check our current and/or resolved issues to find out how we're working with the community to improve persistent storage for our project. For now, a few preparations need to be completed.
Installing Ceph Host Requirements
At some future point, we want to ensure that our solution is cloud-native, allowing installation on any host system without a package manager and only a container runtime (i.e. CoreOS). Until this happens, we will need to ensure that ceph-common is installed on each of our hosts. Using our Ubuntu example:
admin@kubenode01:~$ sudo apt-get install ceph-common -y
We will always attempt to keep host-specific requirements to a minimum, and we are working with the Ceph team (Sébastien Han) to quickly address this Ceph requirement.
Ceph Secrets Generation
Another thing of interest is that our deployment assumes that you can generate secrets at the time of the container deployment. We require the sigil binary on your deployment host in order to perform this action.
admin@kubenode01:~$ curl -L https://github.com/gliderlabs/sigil/releases/download/v0.4.0/sigil_0.4.0_Linux_x86_64.tgz | tar -zxC /usr/local/bin
Kubernetes Controller Manager
Before deploying Ceph, you will need to re-deploy a custom Kubernetes Controller with the necessary RDB utilities. For your convenience, we are maintaining this along with the Openstack-Helm project. If you would like to check the current tags or the security of these pre-built containers, you may view them at our public Quay container registry. If you would prefer to build this container yourself, or add any additional packages, you are free to use our GitHub dockerfiles repository to do so.
To make these changes, export your Kubernetes version, and edit the image line of your kube-controller-manager json manifest on your Kubernetes Master:
admin@kubenode01:~$ export kube_version=v1.5.1
admin@kubenode01:~$ sed -i "s|gcr.io/google_containers/kube-controller-manager-amd64:'$kube_version'|quay.io/attcomdev/kube-controller-manager:'$kube_version'|g" /etc/kubernetes/manifests/kube-controller-manager.json
Now you will want to restart your Kubernetes master server to continue.
Kube Controller Manager DNS Resolution
Until the following Kubernetes Pull Request is merged, you will need to allow the Kubernetes Controller to use the internal container skydns endpoint as a DNS server, and add the Kubernetes search suffix into the controller's resolv.conf. As of now, the Kubernetes controller only mirrors the host's resolv.conf. This is not sufficient if you want the controller to know how to correctly resolve container service endpoints (in the case of DaemonSets).
First, find out what the IP Address of your kube-dns deployment is:
admin@kubenode01:~$ kubectl get svc kube-dns --namespace=kube-system
NAME CLUSTER-IP EXTERNAL-IP PORT(S) AGE
kube-dns 10.96.0.10 <none> 53/UDP,53/TCP 1d
admin@kubenode01:~$
As you can see by this example, 10.96.0.10 is the CLUSTER-IPIP. Now, have a look at the current kube-controller-manager-kubenode01 /etc/resolv.conf:
admin@kubenode01:~$ kubectl exec kube-controller-manager-kubenode01 -n kube-system -- cat /etc/resolv.conf
# Dynamic resolv.conf(5) file for glibc resolver(3) generated by resolvconf(8)
# DO NOT EDIT THIS FILE BY HAND -- YOUR CHANGES WILL BE OVERWRITTEN
nameserver 192.168.1.70
nameserver 8.8.8.8
search jinkit.com
admin@kubenode01:~$
What we need is for kube-controller-manager-kubenode01 /etc/resolv.conf to look like this:
admin@kubenode01:~$ kubectl exec kube-controller-manager-kubenode01 -n kube-system -- cat /etc/resolv.conf
nameserver 10.96.0.10
nameserver 192.168.1.70
nameserver 8.8.8.8
search svc.cluster.local jinkit.com
admin@kubenode01:~$
You can change this by doing the following:
admin@kubenode01:~$ kubectl exec kube-controller-manager-kubenode01 -it -n kube-system -- /bin/bash
root@kubenode01:/# cat <<EOF > /etc/resolv.conf
nameserver 10.96.0.10
nameserver 192.168.1.70
nameserver 8.8.8.8
search svc.cluster.local jinkit.com
EOF
root@kubenode01:/#
Now you can test your changes by deploying a service to your cluster, and resolving this from the controller. As an example, lets deploy something useful, like Kubernetes dashboard:
admin@kubenode01:~$ kubectl create -f https://rawgit.com/kubernetes/dashboard/master/src/deploy/kubernetes-dashboard.yaml
Note the IP field:
admin@kubenode01:~$ kubectl describe svc kubernetes-dashboard -n kube-system
Name: kubernetes-dashboard
Namespace: kube-system
Labels: app=kubernetes-dashboard
Selector: app=kubernetes-dashboard
Type: NodePort
IP: 10.110.207.144
Port: <unset> 80/TCP
NodePort: <unset> 32739/TCP
Endpoints: 10.25.178.65:9090
Session Affinity: None
No events.
admin@kubenode01:~$
Now you should be able to resolve the host kubernetes-dashboard.kube-system.svc.cluster.local:
admin@kubenode01:~$ kubectl exec kube-controller-manager-kubenode01 -it -n kube-system -- ping kubernetes-dashboard.kube-system.svc.cluster.local
PING kubernetes-dashboard.kube-system.svc.cluster.local (10.110.207.144) 56(84) bytes of data.
...
...
admin@kubenode01:~$
(Note: This host example above has iputils-ping installed)
Kubernetes Node DNS Resolution
For each of the nodes to know exactly how to communicate with Ceph (and thus MariaDB) endpoints, each host much also have an entry for kube-dns. Since we are using Ubuntu for our example, place these changes in /etc/network/interfaces to ensure they remain after reboot.
Now we are ready to continue with the Openstack-Helm installation.
Openstack-Helm Preparation
Please ensure that you have verified and completed the steps above to prevent issues with your deployment. Since our goal is to provide a Kubernetes environment with reliable, persistent storage, we will provide some helpful verification steps to ensure you are able to proceed to the next step.
Although Ceph is mentioned throughout this guide, our deployment is flexible to allow you the option of bringing any type of persistent storage. Although most of these verification steps are the same, if not very similar, we will use Ceph as our example throughout this guide.
Node Labels
First, we must label our nodes according to their role. Although we are labeling all nodes, you are free to label only the nodes you wish. You must have at least one, although a minimum of three are recommended. Nodes are labeled according to their Openstack roles:
Storage Nodes: ceph-storage
Control Plane: openstack-control-plane
Compute Nodes: openvswitch, openstack-compute-node
admin@kubenode01:~$ kubectl label nodes openstack-control-plane=enabled --all
admin@kubenode01:~$ kubectl label nodes ceph-storage=enabled --all
admin@kubenode01:~$ kubectl label nodes openvswitch=enabled --all
admin@kubenode01:~$ kubectl label nodes openstack-compute-node=enabled --all
Obtaining the Project
Download the latest copy of Openstack-Helm:
admin@kubenode01:~$ git clone https://github.com/att-comdev/openstack-helm.git
admin@kubenode01:~$ cd openstack-helm
Ceph Preparation and Installation
Ceph must be aware of the OSX cluster and public networks. These CIDR ranges are the exact same ranges you used earlier in your Calico deployment yaml (our example was 10.25.0.0/16 due to our 192.168.0.0/16 overlap). Explore this variable to your deployment environment by issuing the following commands:
admin@kubenode01:~$ export osd_cluster_network=10.25.0.0/16
admin@kubenode01:~$ export osd_public_network=10.25.0.0/16
Ceph Storage Volumes
Ceph must also have volumes to mount on each host labeled for ceph-storage. On each host that you labeled, create the following directory (can be overriden):
admin@kubenode01:~$ mkdir -p /var/lib/openstack-helm/ceph
Repeat this step for each node labeled: ceph-storage
Ceph Secrets Generation
Although you can bring your own secrets, we have conveniently created a secret generation tool for you (for greenfield deployments). You can create secrets for your project by issuing the following:
admin@kubenode01:~$ cd common/utils/secret-generator
admin@kubenode01:~$ ./generate_secrets.sh all `./generate_secrets.sh fsid`
admin@kubenode01:~$ cd ../../..
Nova Compute Instance Storage
Nova Compute requires a place to store instances locally. Each node labeled openstack-compute-node needs to have the following directory:
admin@kubenode01:~$ mkdir -p /var/lib/nova/instances
Repeat this step for each node labeled: openstack-compute-node
Helm Preparation
Now we need to install and prepare Helm, the core of our project. Please use the installation guide from the Kubernetes/Helm repository. Please take note of our required versions above.
Once installed, and initiated (helm init), you will need your local environment to serve helm charts for use. You can do this by:
admin@kubenode01:~$ helm serve . &
admin@kubenode01:~$ helm repo add local http://localhost:8879/charts
Openstack-Helm Installation
Now we are ready to deploy, and verify our Openstack-Helm installation. The first required is to build out the deployment secrets, lint and package each of the charts for the project. Do this my running make in the openstack-helm directory:
admin@kubenode01:~$ make
Helpful Note: If you need to make any changes to the deployment, you may run make again, delete your helm-deployed chart, and redeploy the chart (update). If you need to delete a chart for any reason, do the following:
admin@kubenode01:~$ helm list
NAME REVISION UPDATED STATUS CHART
bootstrap 1 Fri Dec 23 13:37:35 2016 DEPLOYED bootstrap-0.1.0
bootstrap-ceph 1 Fri Dec 23 14:27:51 2016 DEPLOYED bootstrap-0.1.0
ceph 3 Fri Dec 23 14:18:49 2016 DEPLOYED ceph-0.1.0
keystone 1 Fri Dec 23 16:40:56 2016 DEPLOYED keystone-0.1.0
mariadb 1 Fri Dec 23 16:15:29 2016 DEPLOYED mariadb-0.1.0
memcached 1 Fri Dec 23 16:39:15 2016 DEPLOYED memcached-0.1.0
rabbitmq 1 Fri Dec 23 16:40:34 2016 DEPLOYED rabbitmq-0.1.0
admin@kubenode01:~$
admin@kubenode01:~$
admin@kubenode01:~$ helm delete --purge keystone
Please ensure that you use --purge whenever deleting a project.
Ceph Installation and Verification
Install the first service, which is Ceph. If all instructions have been followed as mentioned above, this installation should go smoothly. Use the following command to install Ceph:
admin@kubenode01:~$ helm install --set network.public=$osd_public_network --name=ceph local/ceph --namespace=ceph
Bootstrap Installation
At this time (and before verification of Ceph) you'll need to install the bootstrap chart. The bootstrap chart will install secrets for both the ceph and openstack namespaces for the general StorageClass:
admin@kubenode01:~$ helm install --name=bootstrap-ceph local/bootstrap --namespace=ceph
admin@kubenode01:~$ helm install --name=bootstrap-openstack local/bootstrap --namespace=openstack
You may want to validate that Ceph is deployed successfully. Here are a couple of recommendations for this.
Ceph Validating PVC
To validate persistent volume claim (PVC) creation, we've placed a test manifest in the ./test/ directory. Deploy this pvc and explore the deployment:
admin@kubenode01:~$ kubectl get pvc -o wide --all-namespaces -w
NAMESPACE NAME STATUS VOLUME CAPACITY ACCESSMODES AGE
ceph pvc-test Bound pvc-bc768dea-c93e-11e6-817f-001fc69c26d1 1Gi RWO 9h
admin@kubenode01:~$
The output above indicates that the PVC is 'bound' correctly. Now digging deeper:
admin@kubenode01:~/projects/openstack-helm$ kubectl describe pvc pvc-test -n ceph
Name: pvc-test
Namespace: ceph
StorageClass: general
Status: Bound
Volume: pvc-bc768dea-c93e-11e6-817f-001fc69c26d1
Labels: <none>
Capacity: 1Gi
Access Modes: RWO
No events.
admin@kubenode01:~/projects/openstack-helm$
We can see that we have a VolumeID, and the 'capacity' is 1GB. It is a 'general' storage class. It is just a simple test. You can safely delete this test by issuing the following:
admin@kubenode01:~/projects/openstack-helm$ kubectl delete pvc pvc-test -n ceph
persistentvolumeclaim "pvc-test" deleted
admin@kubenode01:~/projects/openstack-helm$
Ceph Validating StorageClass
Next we can look at the storage class, to make sure that it was created correctly:
admin@kubenode01:~$ kubectl describe storageclass/general
Name: general
IsDefaultClass: No
Annotations: <none>
Provisioner: kubernetes.io/rbd
Parameters: adminId=admin,adminSecretName=pvc-ceph-conf-combined-storageclass,adminSecretNamespace=ceph,monitors=ceph-mon.ceph:6789,pool=rbd,userId=admin,userSecretName=pvc-ceph-client-key
No events.
admin@kubenode01:~$
The parameters is what we're looking for here. If we see parameters passed to the StorageClass correctly, we will see the ceph-mon.ceph:6789 hostname/port, things like userid, and appropriate secrets used for volume claims. This all looks great, and it time to Ceph itself.
Ceph Validation
Most commonly, we want to validate that Ceph is working correctly. This can be done with the following ceph command:
admin@kubenode01:~$ kubectl exec -t -i ceph-mon-0 -n ceph -- ceph status
cluster 046de582-f8ee-4352-9ed4-19de673deba0
health HEALTH_OK
monmap e3: 3 mons at {ceph-mon-392438295-6q04c=10.25.65.131:6789/0,ceph-mon-392438295-ksrb2=10.25.49.196:6789/0,ceph-mon-392438295-l0pzj=10.25.79.193:6789/0}
election epoch 6, quorum 0,1,2 ceph-mon-392438295-ksrb2,ceph-mon-392438295-6q04c,ceph-mon-392438295-l0pzj
fsmap e5: 1/1/1 up {0=mds-ceph-mds-2810413505-gtjgv=up:active}
osdmap e23: 5 osds: 5 up, 5 in
flags sortbitwise
pgmap v22012: 80 pgs, 3 pools, 12712 MB data, 3314 objects
101 GB used, 1973 GB / 2186 GB avail
80 active+clean
admin@kubenode01:~$
Use one of your Ceph Monitors to check the status of the cluster. A couple of things to note above; our health is 'HEALTH_OK', we have 3 mons, we've established a quorum, and we can see that our active mds is 'ceph-mds-2810413505-gtjgv'. We have a healthy environment.
For Glance and Cinder to operate, you will need to create some storage pools for these systems. Additionally, Nova can be configured to use a pool as well, but this is off by default.
kubectl exec -n ceph -it ceph-mon-0 ceph osd pool create volumes 128
kubectl exec -n ceph -it ceph-mon-0 ceph osd pool create images 128
Nova storage would be added like this:
kubectl exec -n ceph -it ceph-mon-0 ceph osd pool create vms 128
The choosing the amount of storage is up to you and can be changed by replacing the 128 to meet your needs.
We are now ready to install our next chart, MariaDB.
MariaDB Installation and Verification
We are using Galera to cluster MariaDB and establish a quorum. To install the MariaDB, issue the following command:
admin@kubenode01:~$ helm install --name=mariadb local/mariadb --namespace=openstack
MariaDB is a StatefulSet (PetSets have been retired in Kubernetes v1.5.0). As such, it initiates a 'seed' which is used to deploy MariaDB members via affinity/anti-affinity features. Ceph uses this as well. So what you will notice is the following behavior:
openstack mariadb-0 0/1 Running 0 28s 10.25.49.199 kubenode05
openstack mariadb-seed-0ckf4 1/1 Running 0 48s 10.25.162.197 kubenode01
NAMESPACE NAME READY STATUS RESTARTS AGE IP NODE
openstack mariadb-0 1/1 Running 0 1m 10.25.49.199 kubenode05
openstack mariadb-1 0/1 Pending 0 0s <none>
openstack mariadb-1 0/1 Pending 0 0s <none> kubenode04
openstack mariadb-1 0/1 ContainerCreating 0 0s <none> kubenode04
openstack mariadb-1 0/1 Running 0 3s 10.25.178.74 kubenode04
What you're seeing is the output of kubectl get pods -o wide --all-namespaces, which is used to monitor the seed host preparing each of the MariaDB/Galera members in order: mariadb-0, then mariadb-1, then mariadb-2. This process can take up to a few minutes, so be patient.
To test MariaDB, do the following:
admin@kubenode01:~/projects/openstack-helm$ kubectl exec mariadb-0 -it -n openstack -- mysql -h mariadb.openstack -uroot -ppassword -e 'show databases;'
+--------------------+
| Database |
+--------------------+
| information_schema |
| keystone |
| mysql |
| performance_schema |
+--------------------+
admin@kubenode01:~/projects/openstack-helm$
Now you can see that MariaDB is loaded, with databases intact! If you're at this point, the rest of the installation is easy. You can run the following to check on Galera:
admin@kubenode01:~/projects/openstack-helm$ kubectl describe po/mariadb-0 -n openstack
Name: mariadb-0
Namespace: openstack
Node: kubenode05/192.168.3.25
Start Time: Fri, 23 Dec 2016 16:15:49 -0500
Labels: app=mariadb
galera=enabled
Status: Running
IP: 10.25.49.199
Controllers: StatefulSet/mariadb
...
...
...
FirstSeen LastSeen Count From SubObjectPath Type Reason Message
--------- -------- ----- ---- ------------- -------- ------ -------
5s 5s 1 {default-scheduler } Normal Scheduled Successfully assigned mariadb-0 to kubenode05
3s 3s 1 {kubelet kubenode05} spec.containers{mariadb} Normal Pulling pulling image "quay.io/stackanetes/stackanetes-mariadb:newton"
2s 2s 1 {kubelet kubenode05} spec.containers{mariadb} Normal Pulled Successfully pulled image "quay.io/stackanetes/stackanetes-mariadb:newton"
2s 2s 1 {kubelet kubenode05} spec.containers{mariadb} Normal Created Created container with docker id f702bd7c11ef; Security:[seccomp=unconfined]
2s 2s 1 {kubelet kubenode05} spec.containers{mariadb} Normal Started Started container with docker id f702bd7c11ef
So you can see that galera is enabled.
Installation of Other Services
Now you can easily install the other services simply by going in order:
Install Memcached/RabbitMQ:
admin@kubenode01:~$ helm install --name=memcached local/memcached --namespace=openstack
admin@kubenode01:~$ helm install --name=rabbitmq local/rabbitmq --namespace=openstack
Install Keystone:
admin@kubenode01:~$ helm install --name=keystone local/keystone --namespace=openstack
Install Horizon:
admin@kubenode01:~$ helm install --name=horizon local/horizon --namespace=openstack
Install Glance:
admin@kubenode01:~$ helm install --name=glance local/glance --namespace=openstack
Final Checks
Now you can run through your final checks. Wait for all services to come up:
admin@kubenode01:~$ watch kubectl get all --namespace=openstack
Finally, you should now be able to access horizon at http:// using admin/password