This isn't really a guide, but more a reference to my own OpenStack lab and how it is set up using Kolla-ansible
- OpenStack Lab
My lab consists of the following for the purpose of this reference:
- 3x API/control nodes
- Intel Celeron CPU J1900
- Dual Ethernet (my network node has Quad Ethernet, but not required)
- 8GB RAM
- CentOS 7
- 3x Compute nodes
- Intel Core i7-5500U CPU
- Dual Ethernet
- 16GB RAM
- CentOS 7
- 1x Dell 6248 Switch
- 1x Synology NAS
Routing should be set up with some initial networks, feel free to change your networks to be whatever you want.
10.10.10.0/24- Primary network for all non tenant/external networking traffic- This will use
VLAN 100 10.10.10.11-20is used for any control nodes10.10.10.21-30is used for any network nodes10.10.10.31-40is used for any storage nodes10.10.10.101-199is used for any compute nodes10.10.10.200-249is used for any other devices in the network10.10.10.250is used for the VIP for keystone/horizon
- This will use
10.10.11.0/24- External network 1 - Used for floating IP addresses- This will use
VLAN 101
- This will use
10.10.12.0/24- External network 2 - Used for floating IP addresses- This will use
VLAN 102
- This will use
10.10.13.0/24- External network 3 - Used for floating IP addresses- This will use
VLAN 103
- This will use
10.10.14.0/24- Amphora network- This will use
VLAN 104
- This will use
You need to make sure that 10.10.10.0/24 can access 10.10.14.0/24 as this is required for Octavia to be able to provision/deprovision amphora instances
Each node must be resolvable by its DNS name, so you can either run your own DNS server like this https://hub.docker.com/r/doko89/technitium or however you want to handle your DNS (if your router can do it, great)
bond0is the network used for everything except for external network and VLAN tenant trafficbond1is used for external network and VLAN tenant traffic
Configurable in kolla-ansible/config/neutron/ml2_conf.ini
100- This is the VLAN all bond0 interfaces run in101-104- These VLANs are used for external networks, these are networks that Floating IP addresses and Amphora images will use105-900- Tenant VLANs
First step is to prepare all the nodes by installing CentOS 7 as the base OS, just do a minimal installation. Don't install anything else, we will run a script over each node that will set them up for use by our eventual run of kolla-ansible
First step is to configure the network interfaces for each node.
You should update the scripts referenced below with any hostname changes you make to the nodes
Review the scripts in node-scripts/**/setup.sh for each node, configure the IP address section to suit your networking ranges. Adjust accordingly and then copy and run them on each of your nodes. Best to do this via keyboard/mouse/monitor or iDRAC/iLO(if your nodes have it)
By default, the configuration is as follows for each node
control01 - 10.10.10.11network01 - 10.10.10.21storage01 - 10.10.10.31compute01 - 10.10.10.101compute02 - 10.10.10.102compute03 - 10.10.10.103
Make sure that wherever you will run kolla-ansible can SSH to all the nodes.
I set up an NFS share on my NAS to provide the cinder backend, feel free to do whatever you want though and not use NFS. If you do create the NFS share, make sure it is accessible by your nodes.
Note: Having an IP in the same subnet and on the same physical switch would be ideal.
Create 2 shares, one for volumes, one for volume backups
10.10.10.200:/volume1/openstack
10.10.10.200:/volume1/openstack-backups
These need to be set in
kolla-ansible/config/nfs_shareskolla-ansible/globals.yml-cinder_backup_share:
I put together a quick docker image that contains kolla-ansible and ansible in the one image.
There is a wrapper for this called kolla.sh that you can use, it just passes everything through to kolla-ansible except for build and genpwd
You can run ./kolla.sh build to build the docker-image if you want, otherwise follow the guide to get setup with kolla-ansible.
Then the first step is to get the latest globals.yml and passwords.yml and populate the passwords with auto generated ones.
./kolla.sh genpwd
This will populate kolla-ansible/passwords.yml locally, keep this somewhere safe.
NOTE: You only need to run this initially to generate passwords.
The inventory file is located in kolla-ansible/multinode adjust this to suit however you want to deploy the different servers across your nodes.
You should update this with any hostname changes you make to the nodes
kolla-ansible/overrides.yml is where we will configure any overrides for the configuration of our openstack.
Take options from the kolla-ansible/globals.yml and add them here to override.
You can also add property config additions/overrides of particular services directly in kolla-ansible/config/<service>
First thing we need to do is bootstrap the servers, this will install everything needed to run kolla-ansible and all the images it uses.
./kolla.sh bootstrap-servers
Once the bootstrap is done, run pre-checks
./kolla.sh prechecks
Once the pre-checks are done, you can deploy the images (this can take quite some time)
./kolla.sh deploy
Once the deployment is done, you can generate a copy of the admin-openrc.sh file running the following
./kolla.sh post-deploy
This will create kolla-ansible/admin-openrc.sh that you can source to interact with openstack using the CLI
Once deployed, you should create all the external network providers ./openstack-scripts/provider-networks.sh
openstack network create --share --external \
--provider-physical-network physnet1 \
--provider-segment 101 \
--provider-network-type vlan \
extnet-vlan101
openstack subnet create --no-dhcp \
--allocation-pool start=10.10.11.10,end=10.10.11.200 \
--network extnet-vlan101 \
--subnet-range 10.10.11.0/24
--gateway 10.10.11.1 \
extnet-vlan101-subnet
openstack network create --share --external \
--provider-physical-network physnet1 \
--provider-segment 102 \
--provider-network-type vlan \
extnet-vlan102
openstack subnet create --no-dhcp \
--allocation-pool start=10.10.12.10,end=10.10.12.200 \
--network extnet-vlan102 \
--subnet-range 10.10.12.0/24
--gateway 10.10.12.1 \
extnet-vlan102-subnet
openstack network create --share --external \
--provider-physical-network physnet1 \
--provider-segment 103 \
--provider-network-type vlan \
extnet-vlan103
openstack subnet create --no-dhcp \
--allocation-pool start=10.10.13.10,end=10.10.13.200 \
--network extnet-vlan103 \
--subnet-range 10.10.13.0/24
--gateway 10.10.13.1 \
extnet-vlan103-subnet
openstack network create --share --external \
--provider-physical-network physnet1 \
--provider-segment 104 \
--provider-network-type vlan \
extnet-vlan104
openstack subnet create --no-dhcp \
--allocation-pool start=10.10.14.10,end=10.10.14.200 \
--network extnet-vlan104 \
--subnet-range 10.10.14.0/24
--gateway 10.10.14.1 \
extnet-vlan104-subnet
You'll need to set up some image flavors for any instances, you can use ones I've created ./openstack-scripts/flavors.sh
openstack flavor delete 01
openstack flavor create m1.nano --id 01 --ram 256 --disk 15 --vcpus 1
openstack flavor delete 02
openstack flavor create m1.tiny --id 02 --ram 512 --disk 15 --vcpus 1
openstack flavor delete 03
openstack flavor create m1.micro --id 03 --ram 1024 --disk 15 --vcpus 1
openstack flavor delete 04
openstack flavor create m1.small --id 04 --ram 2048 --disk 15 --vcpus 1
openstack flavor delete 05
openstack flavor create m1.medium --id 05 --ram 4096 --disk 15 --vcpus 2
openstack flavor delete 06
openstack flavor create m1.large --id 06 --ram 6144 --disk 15 --vcpus 2
openstack flavor delete 07
openstack flavor create m1.xlarge --id 07 --ram 8192 --disk 15 --vcpus 6
openstack flavor delete 08
openstack flavor create m2.nano --id 08 --ram 256 --disk 15 --vcpus 2
openstack flavor delete 09
openstack flavor create m2.tiny --id 09 --ram 512 --disk 15 --vcpus 2
openstack flavor delete 10
openstack flavor create m2.micro --id 10 --ram 1024 --disk 15 --vcpus 2
openstack flavor delete 11
openstack flavor create m2.small --id 11 --ram 2048 --disk 15 --vcpus 2
openstack flavor delete octavia
openstack flavor create --id octavia --disk 20 --private --ram 1024 --vcpus 2 octavia
This guide has a good runthrough on setting up octavia after initially deploying openstack http://www.panticz.de/openstack/octavia
You can deploy just octavia again once configuration options are changed using
./kolla deploy --tags octavia
You're done, you should be able to access horizon now by visiting http://10.10.10.250/ (or whatever IP you configured as your kolla_internal_vip_address in kolla-ansible/overrides.yml file.
The following were used to help in the initial deployment of my openstack lab