Skip to content

konan-abhi/glance-operator

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

GLANCE-OPERATOR

The glance-operator is an OpenShift Operator built using the Operator Framework for Go. The Operator provides a way to install and manage the OpenStack Glance installation on OpenShift. This Operator is developed using RDO containers for OpenStack.

Getting started

NOTES:

  • The project is in a rapid development phase and not yet intended for production consumption, so instructions are meant for developers.

  • If possible don't run things in your own machine to avoid the risk of affecting the development of your other projects.

Here we'll explain how to get a functiona OpenShift deployment running inside a VM that is running MariaDB, RabbitMQ, KeyStone and Glance services against a Ceph backend.

There are 4 steps:

Prerequisites

There are some tools that will be required through this process, so the first thing we do is install them:

sudo dnf install -y git wget make ansible-core python-pip podman gcc

We'll also need this repository as well as install_yamls:

cd ~
git clone https://github.com/openstack-k8s-operators/install_yamls.git
git clone https://github.com/openstack-k8s-operators/glance-operator.git

OpenShift cluster

There are many ways get an OpenShift cluster, and our recommendation for the time being is to use OpenShift Local (formerly known as CRC / Code Ready Containers).

To help with the deployment we have companion development tools available that will install OpenShift Local for you and will also help with later steps.

Running OpenShift requires a considerable amount of resources, even more when running all the operators and services required for an OpenStack deployment, so make sure that you have enough resources in the machine to run everything.

You will need at least 5 CPUS and 16GB of RAM, preferably more, just for the local OpenShift VM.

You will also need to get your pull-secrets from Red Hat and store it in the machine, for example on your home directory as pull-secret.

cd ~/install_yamls/devsetup
PULL_SECRET=~/pull-secret CPUS=6 MEMORY=20480 make download_tools crc

This will take a while, but once it has completed you'll have an OpenShift cluster ready.

Now you need to set the right environmental variables for the OCP cluster, and you may want to logging to the cluster manually (although the previous step already logs in at the end):

eval $(crc oc-env)

NOTE: When CRC finishes the deployment the oc client is logged in, but the token will eventually expire, in that case we can login again with oc login -u kubeadmin -p 12345678 https://api.crc.testing:6443.

Let's now get the cluster version confirming we have access to it:

oc get clusterversion

If you are running OCP on a different machine you'll need additional steps to access its dashboard from an external system.

Storage

There are 2 kinds of storage we'll need: One for the pods to run, for example for the MariaDB database files, and another for the OpenStack services to use for the VMs.

To create the pod storage we run:

cd ~/install_yamls
make crc_storage

As for the storage for the OpenStack services, at the time of this writing only File and Ceph are supported. The Glance Spec can be used to configure Glance to connect to a Ceph RBD server.

Deploy openstack-k8s-operators

Deploying the podified OpenStack control plane is a 2 step process. First deploying the operators, and then telling the openstack-operator how we want our OpenStack deployment to look like.

Deploying the openstack operator:

cd ~/install_yamls
make openstack

Once all the operator ready we'll see the pod with:

oc get pod -l control-plane=controller-manager

And now we can tell this operator to deploy RabbitMQ, MariaDB, Keystone and Glance with File as a backend:

cd ~/install_yamls
make openstack_deploy

After a bit we can see the 4 operators are running:

oc get pods -l control-plane=controller-manager

And a while later the services will also appear:

oc get pods -l app=mariadb
oc get pods -l app.kubernetes.io/component=rabbitmq
oc get pods -l service=keystone
oc get pods -l service=glance

Configure Clients

We can now see available endpoints and services to confirm that the clients and the Keystone service work as expected:

openstack service list
openstack endpoint list

Upload a glance image:

cd
wget http://download.cirros-cloud.net/0.5.2/cirros-0.5.2-x86_64-disk.img -O cirros.img
openstack image create cirros --container-format=bare --disk-format=qcow2 < cirros.img
openstack image list

Cleanup

To delete the deployed OpenStack we can do:

cd ~/install_yamls
make openstack_deploy_cleanup

Once we've done this we need to recreate the PVs that we created at the start, since some of them will be in failed state:

make crc_storage_cleanup crc_storage

We can now remove the openstack-operator as well:

make openstack_cleanup

ADDITIONAL INFORMATION

NOTE: Run make --help for more information on all potential make targets.

More information about the Makefile can be found via the Kubebuilder Documentation.

For developer specific documentation please refer to the Contributing Guide.

Example: configure Glance with Ceph backend

The Glance spec can be used to configure Glance to connect to a Ceph RBD server.

Create a secret which contains the Cephx key and Ceph configuration file so that the Glance pod created by the operator can mount those files in /etc/ceph.

---
apiVersion: v1
kind: Secret
metadata:
  name: ceph-client-conf
  namespace: openstack
stringData:
  ceph.client.openstack.keyring: |
    [client.openstack]
        key = <secret key>
        caps mgr = "allow *"
        caps mon = "profile rbd"
        caps osd = "profile rbd pool=images"
  ceph.conf: |
    [global]
    fsid = 7a1719e8-9c59-49e2-ae2b-d7eb08c695d4
    mon_host = 10.1.1.2,10.1.1.3,10.1.1.4

The following represents an example of Glance resource that can be used to trigger the service deployment, and enable an RBD backend that points to an external Ceph cluster.

apiVersion: glance.openstack.org/v1beta1
kind: Glance
metadata:
  name: glance
spec:
  serviceUser: glance
  containerImage: quay.io/podified-antelope-centos9/openstack-glance-api:current-podified
  customServiceConfig: |
    [DEFAULT]
    enabled_backends = default_backend:rbd
    [glance_store]
    default_backend = default_backend
    [default_backend]
    rbd_store_ceph_conf = /etc/ceph/ceph.conf
    store_description = "RBD backend"
    rbd_store_pool = images
    rbd_store_user = openstack
  databaseInstance: openstack
  databaseUser: glance
  glanceAPI:
    debug:
      service: false
    preserveJobs: false
    replicas: 1
  secret: osp-secret
  storageClass: ""
  storageRequest: 1G
  extraMounts:
    - name: v1
      region: r1
      extraVol:
        - propagation:
          - Glance
          extraVolType: Ceph
          volumes:
          - name: ceph
            projected:
              sources:
              - secret:
                  name: ceph-client-conf
          mounts:
          - name: ceph
            mountPath: "/etc/ceph"
            readOnly: true

When the service is up and running, it will be possible to interact with the Glance API and upload an image using the Ceph backend.

Example: configure Glance with additional networks

The Glance spec can be used to configure Glance to have the pods being attached to additional networks to e.g. connect to a Ceph RBD server on a dedicated storage network.

Create a network-attachement-definition whic then can be referenced from the Glance API CR.

---
apiVersion: k8s.cni.cncf.io/v1
kind: NetworkAttachmentDefinition
metadata:
  name: storage
  namespace: openstack
spec:
  config: |
    {
      "cniVersion": "0.3.1",
      "name": "storage",
      "type": "macvlan",
      "master": "enp7s0.21",
      "ipam": {
        "type": "whereabouts",
        "range": "172.18.0.0/24",
        "range_start": "172.18.0.50",
        "range_end": "172.18.0.100"
      }
    }

The following represents an example of Glance resource that can be used to trigger the service deployment, and have the service pods attached to the storage network using the above NetworkAttachmentDefinition.

apiVersion: glance.openstack.org/v1beta1
kind: Glance
metadata:
  name: glance
spec:
  ...
  glanceAPI:
    ...
    networkAttachents:
    - storage
...

When the service is up and running, it will now have an additional nic configured for the storage network:

# oc rsh glance-external-api-dfb69b98d-mbw42
Defaulted container "glance-api" out of: glance-api, init (init)
sh-5.1# ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host
       valid_lft forever preferred_lft forever
3: eth0@if298: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UP group default
    link/ether 0a:58:0a:82:01:18 brd ff:ff:ff:ff:ff:ff link-netnsid 0
    inet 10.130.1.24/23 brd 10.130.1.255 scope global eth0
       valid_lft forever preferred_lft forever
    inet6 fe80::4cf2:a3ff:feb0:932/64 scope link
       valid_lft forever preferred_lft forever
4: net1@if26: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default
    link/ether a2:f1:3b:12:fd:be brd ff:ff:ff:ff:ff:ff link-netnsid 0
    inet 172.18.0.52/24 brd 172.18.0.255 scope global net1
       valid_lft forever preferred_lft forever
    inet6 fe80::a0f1:3bff:fe12:fdbe/64 scope link
       valid_lft forever preferred_lft forever

Example: expose Glance to an isolated network

The Glance spec can be used to configure Glance to register e.g. the internal endpoint to an isolated network. MetalLB can be used for this scenario.

As a pre requisite, MetalLB needs to be installed and worker nodes prepared to work as MetalLB nodes to serve the LoadBalancer service.

In this example the following MetalLB IPAddressPool is used:

---
apiVersion: metallb.io/v1beta1
kind: IPAddressPool
metadata:
  name: osp-internalapi
  namespace: metallb-system
spec:
  addresses:
  - 172.17.0.200-172.17.0.210
  autoAssign: false

The following represents an example of Glance resource that can be used to trigger the service deployment, and have the internal glanceAPI endpoint registerd as a MetalLB service using the IPAddressPool osp-internal, request to use the IP 172.17.0.202 as the VIP and the IP is shared with other services.

apiVersion: glance.openstack.org/v1beta1
kind: Glance
metadata:
  name: glance
spec:
  ...
  glanceAPI:
    ...
    override:
      service:
        metadata:
          annotations:
            metallb.universe.tf/address-pool: internalapi
            metallb.universe.tf/allow-shared-ip: internalapi
            metallb.universe.tf/loadBalancerIPs: 172.17.0.202
        spec:
          type: LoadBalancer
    ...
...

The internal glance endpoint gets registered with its service name. This service name needs to resolve to the LoadBalancerIP on the isolated network either by DNS or via /etc/hosts:

# openstack endpoint list -c 'Service Name' -c Interface -c URL --service glance
+--------------+-----------+---------------------------------------------------------------+
| Service Name | Interface | URL                                                           |
+--------------+-----------+---------------------------------------------------------------+
| glance       | internal  | http://glance-internal.openstack.svc:9292                     |
| glance       | public    | http://glance-public-openstack.apps.ostest.test.metalkube.org |
+--------------+-----------+---------------------------------------------------------------+

License

Copyright 2022.

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at

http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.

Releases

No releases published

Packages

No packages published

Languages

  • Go 91.0%
  • Makefile 5.2%
  • Shell 3.0%
  • Dockerfile 0.8%