Linux - Ubuntu 16.04 LTS (Xenial Xerus)¶
Most of the configuration steps described in this document apply to any modern Linux environment, however some of them will apply only to Ubuntu and likely to any Ubuntu-based distribution.
This document is based in Ubuntu 16.04 LTS (Xenial Xerus). Once you have installed it, you should be able to follow the instructions described below. In particular, we are going to use Ubuntu packages that can be found under the repositories main and universe.
Important
Please make sure you have reviewed the requirements before proceeding. Also, you may want to consider setting up the firewall before you start installing the services described below to avoid exposing them to outside access.
Install the dependencies¶
MySQL¶
We strongly recommend using MySQL 5.6 as it’s much better than the previous major release in terms of speed, scalability and user-friendliness. Also, we’ve experienced very good results using forks like Percona Server or MariaDB, so don’t be afraid and use them if you want!
Warning
If you are planning to use MySQL 5.7 please be aware that it has not been fully tested yet and we know at least of one bug that it has not been solved yet!
Ubuntu 16.04 does not include an installation candidate for MySQL 5.6 but we can use Percona Server 5.6 instead. Let’s install it with the following command:
sudo apt install percona-server-server-5.6
During the installation, you will be prompted to set a password for the default administrator user (root). We strongly recommend that you use a strong password and please write it down somewhere safe, you are going to need it later.
Elasticsearch¶
A relatively new search server based on Apache Lucene and developed in Java that has brought AtoM a lot of advanced features, performance and scalability. This is probably the biggest change introduced in AtoM 2.x and we are pleased with the results.
Ubuntu doesn’t provide a package but you can download it directly from the Elasticsearch site if you are unable to download it using the method that follows.
First, make sure that Java is installed. In this example we are going to use OpenJDK but Oracle’s JVM would also work.
sudo apt install openjdk-8-jre-headless software-properties-common
After successfully installing Java, proceed to install Elasticsearch. Download and install the public signing key used in their repository:
wget -qO - http://packages.elasticsearch.org/GPG-KEY-elasticsearch | sudo apt-key add -
Now add their repository:
sudo add-apt-repository "deb http://packages.elasticsearch.org/elasticsearch/1.7/debian stable main"
Ready to be installed. Run:
sudo apt update
sudo apt install elasticsearch
Start the service and configure it to start when the system is booted.
sudo systemctl enable elasticsearch
sudo systemctl start elasticsearch
Web server¶
There are many web servers out there capable of working well with PHP. Apache is probably the most popular and we like it, but we’ve found that Nginx adapts itself much better to limited resource environments while it also scales better and more predictably under high loads. You are welcome to try other solutions, but the following documentation will focus merely on Nginx.
Warning
The following instructions assume that the Nginx package is creating the
directory /usr/share/nginx and that is the location where we are
going to place the AtoM sources. However, we have been told this location may
be different in certain environments (e.g. /var/www) or you may opt
for a different location. If that is the case, please make sure that you
update the configuration snippets that we share later in this document
according to your location.
Nginx¶
In Ubuntu, the installation of Nginx is simple:
sudo apt install nginx
Nginx deploys a default server (aka VirtualHost, for Apache users) called
default and you can find it in /etc/nginx/sites-available/default.
In order to install AtoM you could edit the existing server block or add a new
one. We are going to you show you how to do the latter:
sudo touch /etc/nginx/sites-available/atom
sudo ln -sf /etc/nginx/sites-available/atom /etc/nginx/sites-enabled/atom
sudo rm /etc/nginx/sites-enabled/default
We have now created the configuration file and linked it from sites-enabled/, which is the directory that Nginx will look for. This means that you could disable a site by removing its symlink from sites-enabled/ while keeping the original one under sites-available/, in case that you want to re-use it in the future. You can do this with the Nginx default server.
The following is a recommended server block for AtoM. Put the following contents
in /etc/nginx/sites-available/atom.
upstream atom {
server unix:/run/php7.0-fpm.atom.sock;
}
server {
listen 80;
root /usr/share/nginx/atom;
# http://wiki.nginx.org/HttpCoreModule#server_name
# _ means catch any, but it's better if you replace this with your server
# name, e.g. archives.foobar.com
server_name _;
client_max_body_size 72M;
# http://wiki.nginx.org/HttpCoreModule#try_files
location / {
try_files $uri /index.php?$args;
}
location ~ /\. {
deny all;
return 404;
}
location ~* (\.yml|\.ini|\.tmpl)$ {
deny all;
return 404;
}
location ~* /(?:uploads|files)/.*\.php$ {
deny all;
return 404;
}
location ~* /uploads/r/(.*)/conf/ {
}
location ~* ^/uploads/r/(.*)$ {
include /etc/nginx/fastcgi_params;
set $index /index.php;
fastcgi_param SCRIPT_FILENAME $document_root$index;
fastcgi_param SCRIPT_NAME $index;
fastcgi_pass atom;
}
location ~ ^/private/(.*)$ {
internal;
alias /usr/share/nginx/atom/$1;
}
location ~ ^/(index|qubit_dev)\.php(/|$) {
include /etc/nginx/fastcgi_params;
fastcgi_param SCRIPT_FILENAME $document_root$fastcgi_script_name;
fastcgi_split_path_info ^(.+\.php)(/.*)$;
fastcgi_pass atom;
}
location ~* \.php$ {
deny all;
return 404;
}
}
Now you need to enable and reload Nginx:
sudo systemctl enable nginx
sudo systemctl reload nginx
PHP¶
Ubuntu 16.04 bundles PHP 7.0 which is much faster than older releases.
Our favorite way to deploy AtoM is using PHP-FPM, a process manager that scales better than other solutions like FastCGI. The following command will install it along with the rest of PHP extensions required by AtoM:
sudo apt install php7.0-cli php7.0-curl php7.0-json php7.0-ldap php7.0-mysql php7.0-opcache php7.0-readline php7.0-xml php7.0-fpm php7.0-mbstring php7.0-mcrypt php7.0-xsl php7.0-zip php-memcache php-apcu
We also need php-apcu-bc, which is not available yet in Ubuntu 16.04. Let’s
install it manually for now:
sudo apt install php-dev
sudo pecl install apcu_bc-beta
echo "extension=apc.so" | sudo tee /etc/php/7.0/mods-available/apcu-bc.ini
sudo ln -sf /etc/php/7.0/mods-available/apcu-bc.ini /etc/php/7.0/fpm/conf.d/30-apcu-bc.ini
sudo ln -sf /etc/php/7.0/mods-available/apcu-bc.ini /etc/php/7.0/cli/conf.d/30-apcu-bc.ini
sudo systemctl restart php7.0-fpm
Let’s add a new PHP pool for AtoM by adding the following contents in a new file
called /etc/php/7.0/fpm/pool.d/atom.conf:
[atom]
; The user running the application
user = www-data
group = www-data
; Use UNIX sockets if Nginx and PHP-FPM are running in the same machine
listen = /run/php7.0-fpm.atom.sock
listen.owner = www-data
listen.group = www-data
listen.mode = 0600
; The following directives should be tweaked based in your hardware resources
pm = dynamic
pm.max_children = 30
pm.start_servers = 10
pm.min_spare_servers = 10
pm.max_spare_servers = 10
pm.max_requests = 200
chdir = /
; Some defaults for your PHP production environment
; A full list here: http://www.php.net/manual/en/ini.list.php
php_admin_value[expose_php] = off
php_admin_value[allow_url_fopen] = on
php_admin_value[memory_limit] = 512M
php_admin_value[max_execution_time] = 120
php_admin_value[post_max_size] = 72M
php_admin_value[upload_max_filesize] = 64M
php_admin_value[max_file_uploads] = 10
php_admin_value[cgi.fix_pathinfo] = 0
php_admin_value[display_errors] = off
php_admin_value[display_startup_errors] = off
php_admin_value[html_errors] = off
php_admin_value[session.use_only_cookies] = 0
; APC
php_admin_value[apc.enabled] = 1
php_admin_value[apc.shm_size] = 64M
php_admin_value[apc.num_files_hint] = 5000
php_admin_value[apc.stat] = 0
; Zend OPcache
php_admin_value[opcache.enable] = 1
php_admin_value[opcache.memory_consumption] = 192
php_admin_value[opcache.interned_strings_buffer] = 16
php_admin_value[opcache.max_accelerated_files] = 4000
php_admin_value[opcache.validate_timestamps] = 0
php_admin_value[opcache.fast_shutdown] = 1
; This is a good place to define some environment variables, e.g. use
; ATOM_DEBUG_IP to define a list of IP addresses with full access to the
; debug frontend or ATOM_READ_ONLY if you want AtoM to prevent
; authenticated users
env[ATOM_DEBUG_IP] = "10.10.10.10,127.0.0.1"
env[ATOM_READ_ONLY] = "off"
The process manager has to be enabled and restarted:
sudo systemctl enable php7.0-fpm
sudo systemctl start php7.0-fpm
If the service fails to start, make sure that the configuration file has been has been pasted properly. You can also check the syntax by running:
sudo php-fpm7.0 --test
If you are not planning to use the default PHP pool (www), feel free to
remove it:
sudo rm /etc/php/7.0/fpm/pool.d/www.conf
sudo systemctl restart php7.0-fpm
Gearman job server¶
Gearman job server is required by AtoM as of version 2.2.
sudo apt install gearman-job-server
Other packages¶
In order to generate PDF finding aids, AtoM requires Apache FOP 2.1 to be installed. Fortunately, Apache FOP 2.1 can now be installed directly from Ubuntu packages using the command below.
Note
The command specified below uses the --no-install-recommends parameter:
this is intentional and ensures that only dependencies are installed and not
‘recommended’ or ‘suggested’ packages. If --no-install-recommends is not
specified, openjdk-8-jre will be installed as a dependency for one of the
recommended packages. Since openjdk-8-jre-headless was already installed in
the Elasticsearch installation section above, we want to prevent installing
the openjdk-8-jre package as well.
sudo apt install --no-install-recommends fop libsaxon-java
If you want AtoM to be able to process digital objects in formats like JPEG or to extract the text from your PDF documents, there are certain packages that you need to install. They are not mandatory but if they are found in the system, AtoM will use them to produce digital object derivatives from your master objects. for more information on each, see: Requirements: other dependencies. The following will install all the recommended dependencies at once:
sudo apt install imagemagick ghostscript poppler-utils ffmpeg
Download AtoM¶
Now that we have installed and configured all dependencies, we are ready to download and install AtoM itself. The safest way is to install AtoM from the tarball, which you can find in the download section. However, experienced users may prefer to check out the code from our public repository.
The following instructions assume that we are installing AtoM under
/usr/share/nginx and that you are using AtoM 2.3.1.
Option 1: Download the tarball¶
wget https://storage.accesstomemory.org/releases/atom-2.3.1.tar.gz
sudo mkdir /usr/share/nginx/atom
sudo tar xzf atom-2.3.1.tar.gz -C /usr/share/nginx/atom --strip 1
Please note that the tarball may not be available yet if this version is still in development. In this case, you can try the alternative installation method explained below.
Option 2: Check out the code from our git repository¶
Install git:
sudo apt install git
sudo mkdir /usr/share/nginx/atom
sudo git clone -b stable/2.3.x http://github.com/artefactual/atom.git /usr/share/nginx/atom
cd /usr/share/nginx/atom
If you are not interested in downloading all the history from git, you could also truncate it to a specific number of revisions, e.g.: just one revision
git clone --depth 1 http://github.com/artefactual/atom.git /usr/share/nginx/atom
Once you’ve cloned the code from our git repository, you’ll need to compile the CSS files:
curl -sL https://deb.nodesource.com/setup_6.x | sudo -E bash -
sudo apt install nodejs make
sudo npm install -g "less@<2.0.0"
sudo make -C /usr/share/nginx/atom/plugins/arDominionPlugin
Filesystem permissions¶
By default, Nginx runs as the www-data user. There are a few directories under AtoM that must be writable by the web server. The easiest way to ensure this is to update the owner of the AtoM directory and its contents by running:
sudo chown -R www-data:www-data /usr/share/nginx/atom
If you are deploying AtoM in a shared environment we recommend you to pay attention to the permissions assigned to others. The following is an example on how to clear all mode bits for others:
sudo chmod o= /usr/share/nginx/atom
Create the database¶
Assuming that you are running MySQL in localhost, please create the database by running the following command using the password you created earlier:
mysql -h localhost -u root -p -e "CREATE DATABASE atom CHARACTER SET utf8 COLLATE utf8_unicode_ci;"
Notice that the database has been called atom. Feel free to change its name.
In case your MySQL server is not the same as your web server, replace “localhost” with the address of your MySQL server.
Warning
Plase make sure that you are using an empty database! Don’t reuse an old database unless it’s empty. You can always drop it by using the DROP DATABASE command and then create it again.
Additionally, it’s always a good idea to create a specific MySQL user for AtoM
to keep things safer. This is how you can create an user called atom with
password 12345 and the permissions needed for the database created above.
mysql -h localhost -u root -p -e "GRANT ALL ON atom.* TO 'atom'@'localhost' IDENTIFIED BY '12345';"
Note that the INDEX, CREATE and ALTER privileges are only necessary
during the installation process or when you are upgrading AtoM to a newer
version. They can be removed from the user once you are finished with the
installation or you can change the user used by AtoM in config.php.
Run the web installer¶
You should now be ready to run the installer. It’s a simple web interface that changes some internal configuration files according to your environment and adds the necessary tables and initial data to the database recently created.
Open your browser and type the URL in the address bar. The URL can greatly change depending on your web server configuration. The URL will usually be something like http://localhost. AtoM will redirect you to the installer automatically.
The installation process consists of a number of steps where you will be asked for configuration details such as the location of your database server. If you have followed this document to the letter, this is how you should fill the following fields:
- Database name:
atom - Database username:
atom - Database password:
12345 - Database host:
localhost - Database port:
3306 - Search host:
localhost - Search port:
9200 - Search index:
atom
Of course, some of these fields will look very different if you are running AtoM in a distributed way, where your services like MySQL or Elasticsearch are running in separate machines.
The rest of the fields can be filled as you need:
- Site title
- Site description
- Site base URL
- Username
- E-mail address
- Password
Tip
You can always change the site title, site description, and Base URL later via Admin > Settings > Site information. See: Site information for more information. The Username, email, and password can also be changed by an administrator after installation via the user interface - see: Edit an existing user.
Deployment of workers¶
Optionally, you can use Gearman to add support for asynchronous tasks like SWORD deposits, managing rights inheritance, and generating finding aids. Check out the following page for further installation details: Asynchronous jobs and worker management.
Configure AtoM via the command-line¶
There are various settings that can only be configured via the command-line - for example, the default timezone of the application. Depending on your local requirements, it may be preferable to configure some of these now. For more information on these settings see: Manage AtoM configuration files.
Security considerations¶
Now that AtoM is installed, please take a moment to read our security section where we will show you how to configure the firewall in Ubuntu and back up AtoM.
We strongly encourage our users to configure a firewall because some of the services configured should not be exposed in the wild, e.g. Elasticsearch was not designed to be accessible from untrusted networks and it’s a common attack vector.