Access the compute service API¶

Requirements¶

For advanced use cases and automation it is necessary to use the Openstack API endpoints directly.

To do this, contact support to make sure you get an API-enabled user account created for you. A federated user account from e.g SWAMID or Dataporten can't be used directly in this case.

Install the Openstack command line client¶

Openstack.org instructions on how to install the client can be found here. We recommend installing the client in a Python virtual environment.

Linux¶

First install the necessary OS packages depending on what distribution you are using. If the python-virtualenvwrapper package is not available you might have to install epel-release first.

Red Hat Enterprise Linux, CentOS or Fedora

yum install python-devel python-pip python-virtualenvwrapper gcc

Ubuntu or Debian

apt-get install python-dev python-pip virtualenvwrapper build-essential

Installing the client using virtualenvwrapper and pip¶

Restart your shell. Create a virtualenv and install the client into it.

mkvirtualenv oscli
pip install --upgrade pip
pip install python-openstackclient python-neutronclient

To stop or start using this virtual Python environment, type

# to activate it
workon oscli
# to exit, when finished
deactivate

Windows¶

Note

These instructions aren't well tested or perhaps even incomplete. Please let us know if you need help or know how to improve them!

To be able to use the Openstack client from Windows you need Python 2.7. After the installation is finished, open a command prompt:

C:
cd C:\Python27\Scripts

Use easy_install to install pip

C:\Python27\Scripts>easy_install pip

Then install python-openstackclient using pip

C:\Python27\Scripts>pip install python-openstackclient

Configuration and credentials¶

Create a openstackrc file with your environment information and API account credentials. On Linux, you could do it like this:

# create a config file
$ cat - >> openstackrc << EOF
<paste of rc-file from template below>
EOF
# edit it to include real credentials
vi openstackrc
# then activate it
source openstackrc

Run a command to see if it works , this command tests authentication only

openstack token issue

Example scripts to create servers¶

This instruction will show you how to start new servers through the API. To simplify we will first create a network in the gui to which we then connect all our servers.

Make sure that you have a working rc-file according to the insctructions here.

Then create a network according to this instruction.

You should also create a key pair as described here if you want to be able to connect to the servers over SSH.

Now it is time to find the ID of the network:

$ neutron net-list

Find the ID if the network that you just created and save it for later.

Now you want to find the image and the flavor for the server you want to create:

$ openstack image list

Find the image ID of the operating system you want to use and save it for later. You can use an image provided by us, one you have uploaded yourself or a snapshot (which could be practical if the servers you want to create should have certin software installed or be configured in a certain way)

$ openstack flavor list

Find the flavor you want to use. You do not need the ID - only the name

Get the name of you key pair:

$ openstack keypair list

and save the name for later.

Create your server. Everyting inside brackets should be replaced with your own information.

$ openstack server create --image <IMAGE-ID> --flavor <FLAVOR-NAME> --network <NETWORK-ID> --key-name <KEYNAME> <INSTANCE-NAME>

If you would like to create several servers you can create a skript like this:

#!/bin/bash
NETID=xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx
IMAGEID=xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx
FLAVOR=b.tiny
KEYNAME="keyname"

while read servername; do
    openstack server create --image $IMAGEID --flavor $FLAVOR --network $NETID --key-name $KEYNAME $servername
done <servers.txt

You also need to put the names of the servers that you want to create in the text-file servers.txt in the same folder as the script one name per line.

server1
server2
server3

If you want to delete the servers you could use the following script:

#!/bin/bash

while read servername; do
    openstack server delete $servername"
done <servers.txt

In order to connect to you servers you need to add security groups and add floating IPs. If you book floating IPs and create security groups in the portal you can use these alternate scripts to add that to you instances on the fly.

You first need to create a server2.txt file with the following content:

server_name_1;floating_IP_1;security_group
server_name_2;floating_IP_2;security_group

You must ensure that the name of the security group exists and that the floating IPs you are using are reserved in your project but unallocated. Once you have that file in place you can use the following script to create the servers with the correct names, security group and a floating IP assigned to it:

#!/bin/bash
NETID=xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx
IMAGEID=xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx
FLAVOR=b.tiny
KEYNAME="keyname"

while read serverdef; do
    IFS=';' read -r -a serverinfo <<< "$serverdef"
    servername=${serverinfo[0]}
    fip=${serverinfo[1]}
    secgroup=${serverinfo[2]}
    echo "openstack server create --image $IMAGEID --flavor $FLAVOR --security-group $secgroup --network $NETID --key
-name $KEYNAME $servername"
    openstack server create --image $IMAGEID --flavor $FLAVOR --security-group $secgroup --network $NETID --key-name
$KEYNAME $servername
    echo "openstack server add floating ip $servername $fip"
    openstack server add floating ip $servername $fip
done <servers2.txt

And in order to delete the servers again - use this altered script:

#!/bin/bash

while read serverdef; do
    IFS=';' read -r -a serverinfo <<< "$serverdef"
    servername=${serverinfo[0]}
    echo "openstack server delete $servername"
    openstack server delete $servername
done <servers2.txt