Category: Cloud Computing

OpenStack Kilo 3 Node Installation (Controller, Network, Compute) on CentOS 7

install openstack on three nodes
In this tutorial we will install OpenStack Kilo release from RDO repository on three nodes (Controller, Network, Compute) based on CentOS 7 operating system using packstack automated script. The following installation utilizes VLAN based internal software network infrastructure for communication between instances.

Environment used:
public network (Floating IP network): 192.168.2.0/24
internal network (on each node): no IP space, physical connection only (eth1)
controller node public IP: 192.168.2.12 (eth0)
network node public IP: 192.168.2.13 (eth0)
compute node public IP: 192.168.2.14 (eth0)
OS version (each node): CentOS Linux release 7.2.1511 (Core)
install openstack on 3 nodes
Read More

OpenStack: Failed to connect to server (code: 1006)

install openstack on three nodes
In OpenStack Dashboard (Horizon), when I launch an Instance and I am trying to connect to the Instance via Dashboard VNC Console – I am getting error: Failed to connect to server (code: 1006). This is most often caused by improper /etc/nova/nova.conf configuration on Compute node, where the Instance is running:

openstack_vnc_failed_to_connect_to_server_code_1006
Read More

OpenStack: How to Manually Delete Orphaned Neutron Port

openstack
From time to time, especially during testing and automated script tenant deployment, my OpenStack cloud installation gets disordered. Situations when I am unable to delete neutron port in Horizon dashboard are pretty common phenomenon, this most likely leads to problem with deleting network which the problematic port belongs to.
In such cases we may need to edit OpenStack database, which is MariaDB, and manually delete orphaned neutron port related record from MariaDB ports table.

Error: Failed to delete port
Error: Failed to delete port: Port d06e19fc-c637-402b-85a2-3519c69a1035 has owner network:router_interface and therefore cannot be deleted directly via the port API.

Read More

OpenStack Liberty All In One Installation on CentOS 7

openstack
In this tutorial we will install OpenStack Liberty release from RDO repository on one node (all-in-one installation) based on CentOS 7 / RHEL 7 using packstack installer script.

In all-in-one configuration all OpenStack nodes (controller node, compute node, network node) are installed on a single machine. This type of configuration can be deployed very easy and fast for testing purposes, also on KVM machines, when no physical hardware is available.

Environment used:
public network (Floating IP network): 192.168.2.0/24
public network gateway: 192.168.2.1
node IP: 192.168.2.32
node OS: CentOS 7.1 64bit OS
node hardware: 2GHz CPU / 4GB RAM / 1 ethernet card
OpenStack version: Liberty
Read More

OpenStack Command-Line Interface Cheat Sheet

openstack
Openstack, besides Horizon GUI Dashboard, can also be configured via command-line interface using commands in Bash. Below we present a list of common and useful commands for your reference.

Note: in OpenStack command-line interface you have to source appropriate keystonerc file in order to be able to invoke OpenStack commands within the specified Project Tenant. Sourcing of keystonerc file imports Tenant User/Admin credentials to environment variables during user session.
Read More

Bash Script for Automated OpenStack Tenant Creation

openstack
During OpenStack deployment and testing you may need to quickly create: project tenants, tenant users, networks, routers, security groups, etc… just to check, if Cloud is working properly.

Instead of arduous clicking all this stuff in Horizon dashboard, you can use the below Bash script, which will quickly create:
multiple project tenants
2 tenant users (admin, member)
tenant networks, sub-networks
routers
router gateways (to existing public network)
tenant ports in routers
security groups (Allow all TCP,UDP,ICMP ingress/egress)
cirros glance image (cirros-0.3.4-x86_64-disk.img)

Download script: tenant_setup.sh
Read More

Add New Compute Node to Existing OpenStack using Packstack

openstack
OpenStack is quite reliable Cloud solution, that provides extensibility and scalability. That means, if the Cloud is running out of resources for new tenants and instances, it can be easily extended with new Hypervisors (Compute nodes) practically on-line.

In this tutorial we will extend existing OpenStack installation (Controller node, Compute node) with new Compute0 node on-line, without shutting down existing nodes. The easiest and fastest way to extend existing Openstack Cloud on-line is to use Packstack, the automated installer script.
add_new_compute_node
Read More

Install OpenStack on KVM – How To Configure KVM for OpenStack

openstack
Typical OpenStack cloud setup consists of more than one node (usually one Controller node and several Compute nodes), which requires lot of physical computers / servers available to perform the installation. This complicates the situation, especially if we need to test OpenStack on many nodes. Fortunately we can use Linux KVM (Kernel-based Virtual Machine) to create OpenStack on virtual nodes and avoid problems with obtaining physical hardware.

In this tutorial we will show you how to install OpenStack on KVM on Fedora 21 based hypervisor. We will run Openstack installation on two KVM based virtual nodes (Controller and Compute).

The most important part in configuring KVM for OpenStack installation is network setup, we also need to secure some resources (VCPUs, RAM, disk space) on KVM Hypervisor to create two virtual nodes.

openstack-kvm-network-diagram
Read More

How to Manually Extend Cinder Volumes Group in OpenStack

openstack
Cinder Volumes group is a block storage based on LVM (Logical Volume Manager) volume group, usually situated in Controller node in OpenStack cloud. It provides persistent block storage for runtime instances.

During typical packstack installation Cinder Volumes group size is determined in answer file (insert file) by following parameter:

CONFIG_CINDER_VOLUMES_SIZE=20G

If the Cinder Volumes group size we chose upon packstack installation turns out to be insufficient to create new volumes, we can attach additional physical volume on Controller node to the existing Cinder Volumes group and extend it with the new physical volume to gain more space.
Read More