How To Build a Web Server Virtual Machine
The goal here is to build a virtual machine that can act as both a local development environment and a production server. While I'm aiming to keep this generally useful to anyone doing PHP web development, there are places where I install tools or make configuration choices that specifically support my projects. Be aware that there will need to be some improvization on your part if you want this guide to work for you.
The basic features of this environment are:
- Minimal Ubuntu 12.04 "Precise" Server as a VirtualBox guest
- Shared directory between the host and guest for code development
- Firewall configured with UFW
- PHP 5.5 stable, compiled from source
- Nginx 1.4 stable, compiled from source
- MySQL 5.6, compiled from source
- SASS and Compass, for developing CSS
- YUI Compressor, for compressing web assets
The instructions in this guide should be very close to the automated Vagrant build provided along with this document. It should be possible for you to build this environment automatically with:
Install VirtualBox from their download page
Install Vagrant v1.1+
Check out this guide repository somewhere convenient
cd \someplace\convenient git clone git://github.com/triplepoint/web_development_vm_how_to.git how_to
Edit the Vagrant configuration, if necessary
/projectsshare point is the project directory to share between your host and the virtual machine. Edit it appropriately to point to your projects.
web_server_config.vm.networkconfiguration sets the IP address of the virtual machine. If this conflicts for you, you'll need to change it.
Build the machine
Vagrant will use the
makefilepackaged along with this guide to build the environment automatically.
cd \someplace\convenient\how_to\Vagrant vagrant up
Finally, you can get the SSH credentials from Vagrant and use them with whatever SSH client you want
The rest of this guide documents the manual way to build this environment. While there are likely a few differences between the resulting machines, hopefully they're simple cosmetic differences.
On the Host
Create the Guest
Install VirtualBox from their download page
Configure the host-only network
- Start up the VirtualBox Manager
- Go to
- Ensure the existence of (or create) a Host-Only Network with these properties
VirtualBox Host-Only Ethernet Adapter
- IPv4 Address
- IPv4 Network Mask
- Disabled DHCP Server
Download the Ubuntu 12.04 server minimal ISO
From the host's command line, create and configure the new virtual machine (be careful to modify the commands appropriately for suitable names and file paths):
# Create the new disk - make sure the filename path and VM name are modified appropriately VboxManage createhd --filename "E:\Users\username\VirtualBox VMs\new_vm_name\new_vm_name.vdi" --size 10240 --format VDI --variant Standard # Create the new vm - make sure the VM name is modified appropriately VboxManage createvm --name "new_vm_name" --ostype Ubuntu_64 --register VboxManage modifyvm "new_vm_name" --memory 2048 --vram 12 --acpi on --ioapic on --cpus 1 --rtcuseutc on --boot1 dvd --boot2 disk --boot3 none --boot4 none --audio none VboxManage modifyvm "new_vm_name" --nic1 nat --nictype1 82540EM VboxManage modifyvm "new_vm_name" --nic2 hostonly --nictype2 82540EM --hostonlyadapter2 "VirtualBox Host-Only Ethernet Adapter" # Create the HD controller and attach the disk - make sure the medium path and VM name are modified appropriately VBoxManage storagectl "new_vm_name" --name "SATA Controller" --add sata --controller IntelAHCI VBoxManage storageattach "new_vm_name" --storagectl "SATA Controller" --port 0 --device 0 --type hdd --medium "E:\Users\username\VirtualBox VMs\new_vm_name\new_vm_name.vdi" # Create the DVD drive controller and attach the CD ROM ISO - make sure the medium path is where you downloaded the ISO above, and that the VM name is modified appropriately VBoxManage storagectl "new_vm_name" --name "IDE Controller" --add ide VBoxManage storageattach "new_vm_name" --storagectl "IDE Controller" --port 0 --device 0 --type dvddrive --medium "E:\Users\username\Downloads\mini.iso" # Shared folders - verify the VM name, and pick an appropriate hostpath directory to share between the host and guest VBoxManage sharedfolder add "new_vm_name" --name "shared_workspace" --hostpath "E:\Users\username\workspace"
start the virtual machine with:
VBoxManage startvm new_vm_name
Follow all the onscreen Ubuntu setup, accepting defaults. When it comes to selecting packages, select only the
Choose an IP address for the guest (I chose
192.168.56.11below) and set up the Windows hosts file by editing
C:\Windows\System32\drivers\etc\hostsand adding (note these domains are specific to my configuration, yours are likely different):
# Development VM 192.168.56.11 jonathan-hanson.local 192.168.56.11 www.jonathan-hanson.local
If you intend to use IPv6, find the IPv6 zone ID for the VirtualBox Host-Only network by doing the following and reading the 'IDx' column for the 'VirtualBox Host-Only Network':
netsh interface ipv6 show interfaces
Then, (let's say it was 37) in
C:\Windows\System32\drivers\etc\hostsyou can append the zone ID to the virtual host's IPv6 IP address (see below) and add lines like:
At this point it might be worth while to create a backup of the guest's virtual disk to enable future cloning and rollbacks. See the VirtualBox Media Manager for details on how to do this.
On the Guest
Until the network interfaces are set up correctly, you'll need to do this part from the VirtualBox guest directly (that is, not over SSH).
Set Up the Network Interfaces
I noted from
loadapters are present at this point, but
eth1isn't. I did
ifconfig eth1 upand it came up, but with only an ipv6 address, which is incorrect.
It turns out both adapters were configured for DHCP, but the Virtualbox host-only DHCP server is disabled (see above).
eth1with a static IP by adding this to
# The host-only virtualbox interface auto eth1 iface eth1 inet static address 192.168.56.11 netmask 255.255.255.0 network 192.168.56.0 broadcast 192.168.56.255 iface eth1 inet6 static address fe80::0038:000B netmask 64
Reboot the machine and verify that
eth1with the ip address chosen above
At this point you should be able to ssh into the the guest from the host using the IP address chosen above (in my case,
192.168.56.11). On subsequent VM startups you should be able to start it headless with:
vboxmanage startvm SomeVMName --type=headless
Update the Software Packages
Later, when we install tools and dependencies which we've chosen not to build from source, this will ensure we have the latest knowledge of available packages.
Set Up the Firewall
This part is quick and easy.
ufw default deny ufw allow ssh ufw allow http ufw allow 443 ufw enable
Add the VirtualBox Shared Mount
We'll be editing code on the host machine and sharing it to the guest VM with a shared mount.
Create the directory which will be the mount point for the share
Configure the mount by adding to
# virtualbox shared workspace, owned by www-data:www-data shared_workspace /media/sf_shared_workspace vboxsf defaults,uid=33,gid=33 0 0
Mount the shared disk
Set Up the Web Root Symbolic Link
ln -s /media/sf_shared_workspace /var/www
Git is used later by composer.phar, and also to fetch these configuration files.
apt-get install git-core git config --global user.name "Your Name Here" git config --global user.email "email@example.com"
If you're using GitHub, it's likely you'll want to set up an SSH key for this machine. For more information, see GitHub's documentation on SSH keys.
Fetch This Documentation
This documentation includes useful configuration scripts, so we'll fetch it into the VM for easy access:
cd /usr/src/ git clone git://github.com/triplepoint/web_development_vm_how_to.git
Fetch, make, and install:
cd /usr/src/ apt-get install libc6 libpcre3 libpcre3-dev libpcrecpp0 libssl0.9.8 libssl-dev zlib1g zlib1g-dev lsb-base wget http://nginx.org/download/nginx-1.4.2.tar.gz tar -xvf nginx-1.4.2.tar.gz cd nginx-1.4.2 ./configure --prefix=/usr --sbin-path=/usr/sbin --pid-path=/var/run/nginx.pid --conf-path=/etc/nginx/nginx.conf --error-log-path=/var/log/nginx/error.log --http-log-path=/var/log/nginx/access.log --user=www-data --group=www-data --with-http_ssl_module --with-http_spdy_module --with-ipv6 make make install
Install the attached nginx-init script:
cp /usr/src/web_development_vm_how_to/etc/init.d/nginx-init /etc/init.d/nginx chmod 755 /etc/init.d/nginx update-rc.d nginx defaults
Create the Nginx default log directory
Install the generic Nginx config files:
mkdir /etc/nginx/sites-available mkdir /etc/nginx/sites-enabled cp /usr/src/web_development_vm_how_to/etc/nginx/nginx.conf /etc/nginx/ cp /usr/src/web_development_vm_how_to/etc/nginx/sites-available/* /etc/nginx/sites-available
service nginx start
Fetch, make, and install. Note that the test command is optional, but good practice:
cd /usr/src/ apt-get install autoconf libxml2 libxml2-dev libcurl3 libcurl4-gnutls-dev libmagic-dev wget http://us3.php.net/get/php-5.5.2.tar.bz2/from/us2.php.net/mirror -O php-5.5.2.tar.bz2 tar -xvf php-5.5.2.tar.bz2 cd php-5.5.2 ./configure --prefix=/usr --sysconfdir=/etc --with-config-file-path=/etc --without-pear --enable-fpm --with-fpm-user=www-data --with-fpm-group=www-data --enable-opcache --enable-mbstring --enable-mbregex --with-mysqli --with-openssl --with-curl --with-zlib make make test make install
Copy the provided php.ini file to the config directory
cp /usr/src/web_development_vm_how_to/etc/php.ini /etc/php.ini
NOTE - This provided php.ini file is a stripped down custom version. The PHP-provided default versions are present in the php source as
php.ini-development. Feel free to copy one of those instead, but note that the date.timezone will need to be set to avoid warnings.
Install the PHP init script
cp sapi/fpm/init.d.php-fpm /etc/init.d/php-fpm chmod 755 /etc/init.d/php-fpm update-rc.d php-fpm defaults
Copy over the PHP-FPM config file:
cp /etc/php-fpm.conf.default /etc/php-fpm.conf
NOTE - this file needs to be modified after it's copied:
- uncomment the pid directive:
pid = run/php-fpm.pid
- uncomment and set the
error_log = /var/log/php-fpm/php-fpm.log
- changed the
listen = /tmp/php.socket
- uncomment the pid directive:
Create the PHP-FPM and PHP log directories:
mkdir /var/log/php-fpm mkdir /var/log/php
Install the PECL extensions: HTTP, XDebug
pecl update-channels # when prompted, answer with defaults pecl install pecl_http xdebug
/etc/php.ini(skip this step if you're using the custom php.ini provided above):
extension = http.so zend_extension = "/usr/lib/php/extensions/no-debug-non-zts-20121212/xdebug.so"
service php-fpm start
Add the MySQL User
groupadd mysql useradd -c "MySQL Server" -r -g mysql mysql
apt-get install build-essential cmake libaio-dev libncurses5-dev wget http://cdn.mysql.com/Downloads/MySQL-5.6/mysql-5.6.12.tar.gz -O mysql-5.6.12.tar.gz tar -xvf mysql-5.6.12.tar.gz cd mysql-5.6.12 mkdir build && cd build cmake -DCMAKE_INSTALL_PREFIX=/usr/share/mysql -DSYSCONFDIR=/etc .. make make install
Set up the system tables
chown -R mysql:mysql /usr/share/mysql cd /usr/share/mysql/ && scripts/mysql_install_db --user=mysql chown -R root /usr/share/mysql chown -R mysql /usr/share/mysql/data
Set up the MySQL config file
cp /usr/share/mysql/support-files/my-default.cnf /etc/my.cnf
Set up the MySQL init script
cp /usr/share/mysql/support-files/mysql.server /etc/init.d/mysqld chmod 755 /etc/init.d/mysqld update-rc.d mysqld defaults
service mysqld start
Install YUI Compressor
We could fetch this from GitHub and build it manually, but I don't think it's worth the effort. The version in the repository works fine:
apt-get install yui-compressor
We could install Ruby and use Gems ot install compass, but honestly the version in the package repo is just as good.
apt-get install ruby-compass
Periodically it may be necessary to upgrade this machine without rebuilding it. Here's how:
Apt Repository update (covers MySQL, compass, and yui-compressor):
apt-get update; apt-get upgrade;
make cleanand recompile as during the install above
PECL and PEAR
pecl update-channels pecl upgrade pear update-channels pear upgrade
make cleanand recompile as during the install above
Once all upgrades are complete, the various services will need to be restarted
service mysql restart service php-fpm restart service nginx restart
Setting up projects
This guide is primarily aimed at setting up a generic server, and project-specific configuration is probably too specific to get into here. However, here's a few bits and pieces that may be useful in setting up a project.
There's a project makefile example attached to this documenation at
project.makefile.example. It would be appropriate to copy this file to something like
install\makefile in your project code and edit it to make it appropriate for your project. After that you can run:
What follows below is more or less the manual version of what the example makefile above would do.
Nginx config file
There are example nginx config files available with this documentation in
etc/nginx/sites-available. It would be appropriate to copy one of these into /etc/nginx/sites-available with your project's name, and then edit it for your specific project. AFter that you could symlink it into
sites-enabled like so:
cp etc/nginx/sites-available/example /etc/nginx/sites-available/my_project_name ### edit /etc/nginx/sites-available/my_project_name to make it suitable for your project ln -s /etc/nginx/sites-available/my_project_name /etc/nginx/sites-enabled/my_project_name
For Development, its appropriate to have self-signed SSL certs. Depending on your project details, you may need more than one cert. This is more of an example than an exact codeblock to be repeated. See adayinthepit.com's article on self-signed SSL certificates:
Generate the certificate files
mkdir /usr/src/certwork cd /usr/src/certwork openssl genrsa -des3 -out project_name.key 4096 # Enter a password to protect this key openssl req -new -key project_name.key -out project_name.csr # Enter the password from the key above # Answer the questions appropriately (ex, 'US', 'California', 'San Francisco', 'No Company', 'No Org', '*.local_server_name.local', 'firstname.lastname@example.org', '', '' ) # Note that the common name should be the domain you intend to access (ie, '*.local_server_name.local') # Note to leave the password blank. openssl rsa -in project_name.key -out project_name.nginx.key # Enter the password from the key above openssl x509 -req -days 3650 -in project_name.csr -signkey project_name.nginx.key -out project_name.nginx.crt cp project_name.nginx.crt /etc/ssl/certs/ cp project_name.nginx.key /etc/ssl/private/
Be sure to set the cert and key locations in your project's Nginx config file (see above).
Remember that any time you modify nginx config you need to restart it:
service nginx restart
- mysql config
eth1interface doesn't have a static IPV6 address. Need to research how to do that.
- on server errors, Nginx just throws ugly 500 response