Wednesday, September 30, 2015

Automation for DBA - Ansible part 1


In this post I would like to move forward with software provisioning and configuration. In my last post I covered a ways to build a "core VM" and now it's a time to add some software and configuration to it.

There are couple of automation tools, which can be used for that task like Puppet, Chef or Ansible to name a few. The latter one - Ansible - is my favorite, cause in my opinion it has shortest learning curve and also doesn't require any agents on the remote servers.
Although there are some requirements which has to be met on the remote servers:
- ssh connectivity to the remote host
- Python 2.4 ( with python-simplejson ) or 2.5 installed on the remote host.

If Python is a "big no-no" in your organization, you can still you Ansible to help you run a script over set of servers but you will loose almost all functionality.

How Ansible works:

Ansible has a concept of the control machine and a remote managed servers. Ansible scripts are started on the control machine and other servers are managed over a SSH connection and Ansible modules send to the remote servers on demand. Ansible can run in the fully automated mode only if control machine and remote servers has a password-less SSH configuration. In the other case Ansible can ask for a SSH password.

The control machine can be a dedicated sever or it can be an OS admin or DBA laptop, where Ansible is installed. Please check Ansible documentation to find a information how to install it. 

Remote servers can be managed from control machine, only if they are added into a Ansible inventory. The inventory is a text file with the following format:

envtest ansible_ssh_host= ansible_ssh_user=oracle


There are two groups in inventory file - testdbgroup with one members envtest, and proddb-group with two members rac1 and rac2. Server - testdb - is listed in the inventory file but it not a part of any group. 

Ansible has a concept of host and group variables. 
Group variables are used for all host defined in the a group. There is a directory called group_vars, containing a files named with a group name. Each file is a set of variables for a particular group.
Host variables are used for a particular host only and they overwrite a group variables. There is a directory called host_vars, containing a files named with a host name. Each file is a set of variables for a particular host.

$ ls -l host_vars/
total 24
-rw-r--r--  1 mprzepiorowski  staff  29 Sep 29 12:44 envtest

$ cat host_vars/envtest
password: "host env password"

$ ls -l group_vars/
total 8
-rw-r--r--  1 mprzepiorowski  staff  23 Sep 29 13:10 testgroup

$ cat group_vars/testgroup
groupname: "Test group"
Server envtest has a file with host variables and it belongs to the group testgroup with a defined variables as well. Both will be used in the following example. 

Running mode
There are three ways of running a remote command:
- single command mode, 
- Ansible playbooks
- Ansible playbooks with roles
and commands or playbooks can be run on the specific host from inventory file, specific group or all hosts.


Single command:

In the first run Ansible can be used to add SSH public key from control machine into remote server. By default Ansible connect to the remote servers using a user running a command on the control machine, but this can be changed in the inventory file.
$ cat inventory/lab
envtest ansible_ssh_host= ansible_ssh_user=oracle
Remote server has a user oracle and it will be used by all scripts. If remote server name "envtest" is not resolved by a DNS, IP address can be added to inventory file as well.

Test a SSH connection to remote host
$ ssh oracle@
oracle@'s password:
Password is required and this is an example command to fix this problem

$ ansible -k -i inventory/lab all -m authorized_key -a "user=oracle key=\"{{ lookup('file', '/Users/mprzepiorowski/.ssh/') }}\" "
SSH password:

envtest | success >> {
    "changed": true,
    "key": "ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAABAQD43SwP3zXR/LrKqADJDC8jUOv0vgMEQVqWOXoUgwlzQ1vpS1I5m3GjXggG7fjU5I2jGbG+V9xSqPZdw4MGed4nsgOo1m68w9oBpfYLu3hKACSw1Tgu8Ghxd5TB9ohyZ7NOgepXB+zmV8NoqvAUg4yDkk4qPn1fQmoIz7yLkPRErnmSkI6e/gHsGAmZ+5WdvH0FByPCMEfHqDe8vI4ZPVbPWZzl5x9m72HtxjCbB2htHg1JEcch6927oiRE9rllbOL0M/tw5LgjCtYpM3iZTsBwFsGZ6NQ1DM+OiJSrDEhUqUMOZ08X/G7aNYb28QUt9G/FZTTwZjEfnOO75n3i29U/ mprzepiorowski@Marcins-MacBook-Pro.local",
    "key_options": null,
    "keyfile": "/home/oracle/.ssh/authorized_keys",
    "manage_dir": true,
    "path": null,
    "state": "present",
    "unique": false,
    "user": "oracle"
Command options:
- k - ask for SSH password
- i inventory/lab - use a inventory file from folder inventory and file name called lab
all - run this command for all hosts from inventory file
-m authorized_key - use Ansible module authorized_key
- a "user=oracle, key=xxxxxxxx" - Ansible module parameters 

Command output:
- changed : true - remote system has been changed due to Ansible call
- envtest | success - action returned OK status
- other lines are depend on the module

Ansible returned OK status so now SSH connection should work without password:
$ ssh oracle@
Last login: Mon Sep 28 15:58:24 2015 from
[oracle@envtest ~]$
Ansible did a work and SSH public key has been added to user configuration on the remote server. 

Single command is a nice feature if you want to run single command across fleet of servers, but for most of cases Ansible playbook will be a way to deal with.

Ansible playbook:

Next example will show how to create a simple playbook with an one task and install a vim package using yum

Playbook file is formatted in YAML and it looks like this:
- hosts: all
     - name: install vim
       sudo: yes
       yum: name=vim state=present
Be aware that YAML file syntax include a white space, so format of this file matter as well.

Line description:
Hosts: all - mean run a playbook for all hosts from inventory file
tasks: - is a start of task lists (one task in this example)
name: - start of task definition
sudo: - run this module using sudo
yum: - module name with parameters (name is a package to install, state=present mean install it)

Ansible playbook can be stated using this command:
$ ansible-playbook yum.yml -i inventory/lab

PLAY [all] ********************************************************************

GATHERING FACTS ***************************************************************
ok: [envtest]

TASK: [install vim] ***********************************************************
changed: [envtest]

PLAY RECAP ********************************************************************
envtest                    : ok=2    changed=1    unreachable=0    failed=0

Where -i is pointing to the inventory file and yum.yml is a name of file with playbook definition. Output contains a list of tasks executed and summary with number of successes, changes and failures. 

Ansible playbook and variables:

In the all previous examples, only has host has been used and there was no variables set. Let's add some more complexity to playbook and show how flexible is Ansible. Both variables defined for particular host and group will be used to create a text file based on template.

Template file ( Ansible supports Jinja2 template language)
$ cat template.j2
Host name is {{ ansible_hostname }}
Host var password is set to {{ password }}
Group var groupname is set to {{ groupname }}

Ansible playbook:
$ cat template.yml
- hosts: all
     - name: generate scripts
       template: src=template.j2 dest=/tmp/output mode=0644

Running playbook:
$ ansible-playbook -i inventory/lab template.yml

PLAY [all] ********************************************************************

GATHERING FACTS ***************************************************************
ok: [envtest]

TASK: [generate scripts] ******************************************************
changed: [envtest]

PLAY RECAP ********************************************************************
envtest                    : ok=2    changed=1    unreachable=0    failed=0
Output file on envtest server
$ ssh oracle@
Last login: Tue Sep 29 12:29:59 2015 from
[oracle@envtest ~]$ cat /tmp/output
Host name is envtest
Host var password is set to host env password
Group var groupname is set to Test group
[oracle@envtest ~]$

If you think that Ansible is difficult at that stage, and there is no use for DBA's wait until next blog post when I will use a ansible-Oracle playbooks to install Oracle on server changing only one files with a variables. 




Sunday, September 20, 2015

Automation for DBA - Vagrant part 2


In the last post I presented steps to create Virtual Machine using Vagrant with Virtual Box. It is a native combination but there are other possibilities as well. This post will cover an integration of Vagrant tool with KVM hypervisor and AWS EC2 cloud.

In both cases, an additional plugins has to be installed:
-    vagrant-libvirt plugin for KVM -
-    vagrant-aws plugin for EC2 -

If there is a requirement to use a Vagrant boxes defined for other hypervisors, there is a plugin called vagrant-mutate, which help you to convert it from one format to another.

KVM installation steps:

Host configuration:
-    Fedora 21
-    QEMU emulator version 2.1.

List of plugins is maintain on this page - Available-Vagrant-Plugins

Vagrant has very easy way to manage plugins. If plugin is listed on the official page installation process is limited to single command.

[pioro@piorovm]$ vagrant plugin install vagrant-mutate
Installing the 'vagrant-mutate' plugin. This can take a few minutes...
Installed the plugin 'vagrant-mutate (1.0.1)'!

[pioro@piorovm]$ vagrant plugin install vagrant-libvirt
Installing the 'vagrant-libvirt' plugin. This can take a few minutes...
Installed the plugin 'vagrant-libvirt (0.0.30)'!
Displaying a list of installed plugins with version.
[pioro@piorovm]$ vagrant plugin list
vagrant-libvirt (0.0.30)
vagrant-mutate (1.0.1)
vagrant-share (1.1.4)
As in the previous post a prebuild box racattack/oracle65 will be used. This box has been defined for Virtual Box so it has to be downloaded and migrated into KVM format. There is one caveat – after conversion new box will be still using a SATA drivers and not a native virtio drivers. If performance of Vagrant box is important, I will recommend creating a new native version of box for KVM configured with virtio.

Box racattack has two versions in official repository and mutate plugin was unable to use it directly. To workaround that issue a box has to be downloaded first using ex. wget. 
[pioro@piorovm ~]$ wget
--2015-08-19 23:33:57--
Resolving (,
Connecting to (||:443... connected.
HTTP request sent, awaiting response... 302 Found
Location: [following]
--2015-08-19 23:33:58--
Resolving (
Connecting to (||:443... connected.
HTTP request sent, awaiting response... 302 FOUND
Location: [following]
--2015-08-19 23:33:58--
Resolving (,,, ...
Connecting to (||:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 1185789020 (1.1G) [application/octet-stream]
Saving to: ‘’
In the next steps it will be renamed, converted and added to box's repository.
[pioro@piorovm ~]$ mv

[pioro@piorovm ~]$ vagrant mutate libvirt
Extracting box file to a temporary directory.
Converting racattack from virtualbox to libvirt.
Cleaning up temporary files.
The box racattack (libvirt) is now ready to use.

[pioro@piorovm ~]$ vagrant box list
homeora65 (libvirt, 0)
precise32 (virtualbox, 0)
racattack (libvirt, 0)
Vagrant configuration is ready and a new Virtual Machine can be created. Like in the previous host a new directory will be created and a Vagrant will be initialized. In the next step a VM will be started.
[pioro@piorovm ~]$ cd test
[pioro@piorovm test]$ vagrant init racattack
A `Vagrantfile` has been placed in this directory. You are now
ready to `vagrant up` your first virtual environment! Please read
the comments in the Vagrantfile as well as documentation on
`` for more information on using Vagrant.
[pioro@piorovm test]$ vagrant up
Bringing machine 'default' up with 'libvirt' provider...
==> default: Uploading base box image as volume into libvirt storage...
==> default: Creating image (snapshot of base box volume).
==> default: Creating domain with the following settings...
==> default:  -- Name:              test_default
==> default:  -- Domain type:       kvm
==> default:  -- Cpus:              1
==> default:  -- Memory:            512M
==> default:  -- Base box:          racattack
==> default:  -- Storage pool:      default
==> default:  -- Image:             /VM/d1/images/test_default.img
==> default:  -- Volume Cache:      default
==> default:  -- Kernel:
==> default:  -- Initrd:
==> default:  -- Graphics Type:     vnc
==> default:  -- Graphics Port:     5900
==> default:  -- Graphics IP:
==> default:  -- Graphics Password: Not defined
==> default:  -- Video Type:        cirrus
==> default:  -- Video VRAM:        9216
==> default:  -- Keymap:            en-us
==> default:  -- Command line :
==> default: Creating shared folders metadata...
==> default: Starting domain.
==> default: Waiting for domain to get an IP address...
==> default: Waiting for SSH to become available...
    default: Vagrant insecure key detected. Vagrant will automatically replace
    default: this with a newly generated keypair for better security.
    default: Inserting generated public key within guest...
    default: Removing insecure key from the guest if its present...
    default: Key inserted! Disconnecting and reconnecting using new SSH key...
==> default: Configuring and enabling network interfaces...
==> default: Rsyncing folder: /home/pioro/test/ => /vagrant
[pioro@piorovm test]$
There is a difference how files are synchronized between Vagrant and a VM machine in VirtualBox and other providers. Virtual Box is using a shared directory feature and KVM and EC2 are using rsync. Rsync process is started only once during VM start, and it to synchronize files after that “vagrant rsync” command has to be run

Like I mentioned in previous post, there are differences in Vagrant configuration between hypervisors. 

The following example is showing a configuration which will add two disks:
 - 10 GB none shared disk and 
 - 20 GB disk which can be shared between hosts. 

Memory size is set to 4 GB and number of vCPUs is set to 2. 
VM will have two network interfaces - one bridged to host network card (192.168.1.x) and one private KVM network (10.10.10.x)

Vagrant.configure(2) do |config| = "racattack "

  config.vm.provider :libvirt do |libvirt|
 libvirt.storage_pool_name = "pool_d1"
        libvirt.cpus=2 :file, :size => '20G', :format => 'qcow2 :file, :size => '10G', :type => 'raw', :allow_existing => 'true', :bus=> 'scsi', :device=>'sda', :path=>'asmdisk1_2'

  config.vm.define :host do |host|
         host.vm.hostname = "targetkvm" :public_network, :dev => "br0", :mode => "bridge", :type => "bridge", :ip => "" :private_network, :libvirt__network_name => "private", :ip => ""

This is all about KVM for now but I believe this is a good starting point for experiments.

EC2 configuration

AWS integration with Vagrnat is based on plugin as well. The main difference is that Vagrant box is limited to a file definition and VM is based on AMI delivered by AWS.
This AMI image can be official AWS build, community one or the one created by you.

Installation of the additional plugin:

$ vagrant plugin install vagrant-aws
Installing the 'vagrant-aws' plugin. This can take a few minutes...
Installed the plugin 'vagrant-aws (0.6.0)'!
List of all plugins
$ vagrant plugin list
vagrant-aws (0.6.0)
vagrant-share (1.1.4, system)
Let's create a new directory and configure an EC2 container to start
$ mkdir awstest
$ cd awstest/
$ vagrant init
In the next step a VM has to be configured. This is example Vagrantfile
Vagrant.configure("2") do |config|

  # for AWS use dummy box = "dummy"            

  config.vm.provider :aws do |aws, override|
    aws.access_key_id = "KEY_ID"
    aws.secret_access_key = "SECRET_ACCESS_KEY"
    aws.instance_type = "t1.micro"
    aws.security_groups = "quick-start-1"
    aws.keypair_name = "mykeys"
    aws.region = "eu-west-1"

    # this is official CentOS AMI
    # agree to CentOS license -
    aws.ami = "ami-42718735"

    override.ssh.username = "root"
    override.ssh.private_key_path = "/tmp/mykey.pem"

    aws.block_device_mapping = [{ 'DeviceName' => '/dev/sdb1', 'Ebs.VolumeSize' => 50 }]
Next step is easy "vagrant up" will start your EC2 instance
$ vagrant up
Bringing machine 'default' up with 'aws' provider...
==> default: Warning! The AWS provider doesn't support any of the Vagrant
==> default: high-level network configurations (``). They
==> default: will be silently ignored.
==> default: Launching an instance with the following settings...
==> default:  -- Type: t1.micro
==> default:  -- AMI: ami-42718735
==> default:  -- Region: eu-west-1
==> default:  -- Keypair: mykeys
==> default:  -- Security Groups: ["quick-start-1"]
==> default:  -- Block Device Mapping: [{"DeviceName"=>"/dev/sdb1", "Ebs.VolumeSize"=>50}]
==> default:  -- Terminate On Shutdown: false
==> default:  -- Monitoring: false
==> default:  -- EBS optimized: false
==> default:  -- Assigning a public IP address in a VPC: false
==> default: Waiting for instance to become "ready"...
==> default: Waiting for SSH to become available...
==> default: Machine is booted and ready for use!
==> default: Rsyncing folder: /tmp/awstest/ => /vagrant
The following SSH command responded with a non-zero exit status.
Vagrant assumes that this means the command failed!

mkdir -p '/vagrant'

Stdout from the command:

Stderr from the command:

sudo: sorry, you must have a tty to run sudo
Instance is up and running. There was a problem with a rsync as there is problem with SUDO configuration in this AMI. It can be resolved but this is not an issue for this post. Let's try to ssh to EC2 VM now
$ vagrant ssh
[root@ip-10-105-167-95 ~]# ls -l /dev/x*
brw-rw----. 1 root disk 202, 64 Sep 20 10:43 /dev/xvde
brw-rw----. 1 root disk 202, 81 Sep 20 10:43 /dev/xvdf1
SSH to EC2 with Vagrant is very simple. You don't need to remember that is a public IP of EC2 instance, "vagrant ssh" will do a trick. 
To stop EC2 instance use "vagrant halt" and to delete EC2 instance use "vagrant destroy"

This is all I would like to share in this post. In the next one, I will use Vagrant with VirtualBox and I will show how to add an automatic Oracle installation to your VM.