LP#1315531 Quick clarification about which Ubuntu we mean.

[working/Evergreen.git] / docs / installation / server_installation.txt

Installing the Evergreen server
===============================
:toc:
:numbered:

Preamble: referenced user accounts
----------------------------------

In subsequent sections, we will refer to a number of different accounts, as
follows:

  * Linux user accounts:
    ** The *user* Linux account is the account that you use to log onto the
       Linux system as a regular user.
    ** The *root* Linux account is an account that has system administrator
       privileges. On Debian and Fedora you can switch to this account from
       your *user* account by issuing the `su -` command and entering the
       password for the *root* account when prompted. On Ubuntu you can switch
       to this account from your *user* account using the `sudo su -` command
       and entering the password for your *user* account when prompted.
    ** The *opensrf* Linux account is an account that you create when installing
       OpenSRF. You can switch to this account from the *root* account by
       issuing the `su - opensrf` command.
    ** The *postgres* Linux account is created automatically when you install
       the PostgreSQL database server. You can switch to this account from the
       *root* account by issuing the `su - postgres` command.
  * PostgreSQL user accounts:
    ** The *evergreen* PostgreSQL account is a superuser account that you will
       create to connect to the PostgreSQL database server.
  * Evergreen administrator account:
    ** The *egadmin* Evergreen account is an administrator account for
       Evergreen that you will use to test connectivity and configure your
       Evergreen instance.

Preamble: developer instructions
--------------------------------

[NOTE]
Skip this section if you are using an official release tarball downloaded
from http://evergreen-ils.org/downloads

Developers working directly with the source code from the Git repository,
rather than an official release tarball, must install some extra packages
and perform one step before they can proceed with the `./configure` step.

As the *root* Linux account, install the following packages:

  * autoconf
  * automake
  * libtool

As the *user* Linux account, issue the following command in the Evergreen
source directory to generate the configure script and Makefiles:

[source, bash]
------------------------------------------------------------------------------
autoreconf -i
------------------------------------------------------------------------------

Installing prerequisites
------------------------

Evergreen has a number of prerequisite packages that must be installed
before you can successfully configure, compile, and install Evergreen.

1. Begin by installing the most recent version of OpenSRF (2.3.0 or later).
   You can download OpenSRF releases from http://evergreen-ils.org/opensrf-downloads/
2. On many distributions, it is necessary to install PostgreSQL 9 from external
   repositories.
+
  * On Debian Squeeze, open `/etc/apt/sources.list` in a text editor as the
    *root* Linux account and add the following line:
+
[source, bash]
------------------------------------------------------------------------------
deb http://backports.debian.org/debian-backports squeeze-backports main contrib
------------------------------------------------------------------------------
+
  * On Ubuntu Lucid, you can use a PPA (personal package archive), which are 
    package sources hosted on Launchpad. The one most commonly used by Evergreen
    Community members is maintained by Martin Pitt, who also maintains the
    official PostgreSQL packages for Ubuntu. As the *root* Linux account, issue
    the following commands to add the PPA source:
+
[source, bash]
------------------------------------------------------------------------------
apt-get install python-software-properties
add-apt-repository ppa:pitti/postgresql
------------------------------------------------------------------------------
+
  * Ubuntu Precise comes with PostgreSQL 9, so no additional steps are required.
+
  * Fedora comes with PostgreSQL 9, so no additional steps are required.
+
3. On Debian and Ubuntu, run `aptitude update` as the *root* Linux account to
   retrieve the new packages from the backports repository.
4. Issue the following commands as the *root* Linux account to install
   prerequisites using the `Makefile.install` prerequisite installer,
   substituting `debian-squeeze`, `debian-wheezy`, `fedora`, `ubuntu-lucid`, or
   `ubuntu-precise` for <osname> below:
+
[source, bash]
------------------------------------------------------------------------------
make -f Open-ILS/src/extras/Makefile.install <osname>
------------------------------------------------------------------------------
+
5. Add the libdbi-libdbd libraries to the system dynamic library path by
   issuing the following commands as the *root* Linux account:
+
[NOTE]
You should skip this step if installing on Ubuntu Precise or Trusty. The ubuntu
targets use libdbd-pgsql from packages.
+
.Debian
[source, bash]
------------------------------------------------------------------------------
echo "/usr/local/lib/dbd" > /etc/ld.so.conf.d/eg.conf
ldconfig
------------------------------------------------------------------------------
+
.Fedora
[source, bash]
------------------------------------------------------------------------------
echo "/usr/lib64/dbd" > /etc/ld.so.conf.d/eg.conf
ldconfig
------------------------------------------------------------------------------

Configuration and compilation instructions
------------------------------------------

For the time being, we are still installing everything in the `/openils/`
directory. From the Evergreen source directory, issue the following commands as
the *user* Linux account to configure and build Evergreen:

[source, bash]
------------------------------------------------------------------------------
PATH=/openils/bin:$PATH ./configure --prefix=/openils --sysconfdir=/openils/conf
make
------------------------------------------------------------------------------

These instructions assume that you have also installed OpenSRF under `/openils/`.
If not, please adjust PATH as needed so that the Evergreen `configure` script
can find `osrf_config`.

Installation instructions
-------------------------

1. Once you have configured and compiled Evergreen, issue the following
   command as the *root* Linux account to install Evergreen, build the server
   portion of the staff client, and copy example configuration files to
   `/openils/conf`.
   Change the value of the `STAFF_CLIENT_STAMP_ID` variable to match the version
   of the staff client that you will use to connect to the Evergreen server.
+
[source, bash]
------------------------------------------------------------------------------
make STAFF_CLIENT_STAMP_ID=rel_name install
------------------------------------------------------------------------------
+
2. The server portion of the staff client expects `http://hostname/xul/server`
   to resolve. Issue the following commands as the *root* Linux account to
   create a symbolic link pointing to the `server` subdirectory of the server
   portion of the staff client that we just built using the staff client ID
   'rel_name':
+
[source, bash]
------------------------------------------------------------------------------
cd /openils/var/web/xul
ln -sf rel_name/server server
------------------------------------------------------------------------------

Change ownership of the Evergreen files
---------------------------------------

All files in the `/openils/` directory and subdirectories must be owned by the
`opensrf` user. Issue the following command as the *root* Linux account to
change the ownership on the files:

[source, bash]
------------------------------------------------------------------------------
chown -R opensrf:opensrf /openils
------------------------------------------------------------------------------

Additional Instructions for Developers
--------------------------------------

[NOTE]
Skip this section if you are using an official release tarball downloaded
from http://evergreen-ils.org/downloads

Developers working directly with the source code from the Git repository,
rather than an official release tarball, need to install the Dojo Toolkit
set of JavaScript libraries. The appropriate version of Dojo is included in
Evergreen release tarballs. Developers should install the Dojo 1.3.3 version
of Dojo by issuing the following commands as the *opensrf* Linux account:

[source, bash]
------------------------------------------------------------------------------
wget http://download.dojotoolkit.org/release-1.3.3/dojo-release-1.3.3.tar.gz
tar -C /openils/var/web/js -xzf dojo-release-1.3.3.tar.gz
cp -r /openils/var/web/js/dojo-release-1.3.3/* /openils/var/web/js/dojo/.
------------------------------------------------------------------------------


Configure the Apache Web server
-------------------------------

1. Use the example configuration files in `Open-ILS/examples/apache/` (for
Apache versions below 2.4) or `Open-ILS/examples/apache_24/` (for Apache
versions 2.4 or greater) to configure your Web server for the Evergreen
catalog, staff client, Web services, and administration interfaces. Issue the
following commands as the *root* Linux account:
+
.Debian and Ubuntu Precise
[source,bash]
------------------------------------------------------------------------------
cp Open-ILS/examples/apache/eg.conf       /etc/apache2/sites-available/
cp Open-ILS/examples/apache/eg_vhost.conf /etc/apache2/
cp Open-ILS/examples/apache/eg_startup    /etc/apache2/
# Now set up SSL
mkdir /etc/apache2/ssl
cd /etc/apache2/ssl
------------------------------------------------------------------------------
+
.Ubuntu Trusty
[source,bash]
------------------------------------------------------------------------------------
cp Open-ILS/examples/apache_24/eg_24.conf       /etc/apache2/sites-available/eg.conf
cp Open-ILS/examples/apache_24/eg_vhost_24.conf /etc/apache2/eg_vhost.conf
cp Open-ILS/examples/apache/eg_startup          /etc/apache2/
# Now set up SSL
mkdir /etc/apache2/ssl
cd /etc/apache2/ssl
------------------------------------------------------------------------------------
+
.Fedora
[source,bash]
------------------------------------------------------------------------------
cp Open-ILS/examples/apache_24/eg_24.conf       /etc/httpd/conf.d/
cp Open-ILS/examples/apache_24/eg_vhost_24.conf /etc/httpd/eg_vhost.conf
cp Open-ILS/examples/apache/eg_startup          /etc/httpd/
# Now set up SSL
mkdir /etc/httpd/ssl
cd /etc/httpd/ssl
------------------------------------------------------------------------------
+
2. The `openssl` command cuts a new SSL key for your Apache server. For a
production server, you should purchase a signed SSL certificate, but you can
just use a self-signed certificate and accept the warnings in the staff client
and browser during testing and development. Create an SSL key for the Apache
server by issuing the following command as the *root* Linux account:
+
[source,bash]
------------------------------------------------------------------------------
openssl req -new -x509 -days 365 -nodes -out server.crt -keyout server.key
------------------------------------------------------------------------------
+
3. As the *root* Linux account, edit the `eg.conf` file that you copied into
place.
  a. To enable access to the offline upload / execute interface from any
     workstation on any network, make the following change (and note that
     you *must* secure this for a production instance):
     * (Apache 2.2): Replace `Allow from 10.0.0.0/8` with `Allow from all`
     * (Apache 2.4): Replace `Require host 10.0.0.0/8` with `Require all granted`
  b. (Fedora): Change references from the non-existent `/etc/apache2/` directory
     to `/etc/httpd/`.
4. Change the user for the Apache server.
  * (Debian and Ubuntu): As the *root* Linux account, edit
    `/etc/apache2/envvars`.  Change `export APACHE_RUN_USER=www-data` to 
    `export APACHE_RUN_USER=opensrf`.
  * (Fedora): As the *root* Linux account , edit `/etc/httpd/conf/httpd.conf`.
    Change `User apache` to `User opensrf`.
5. Configure Apache with performance settings appropriate for Evergreen:
  * (Debian and Ubuntu): As the *root* Linux account, edit
    `/etc/apache2/apache2.conf`:
  * (Fedora): As the *root* Linux account, edit `/etc/httpd/conf/httpd.conf`:
    a. Change `KeepAliveTimeout` to `1`. Higher values reduce the chance of
       a request timing out unexpectedly, but increase the risk of using up
       all available Apache child processes.
    b. 'Optional': Change `MaxKeepAliveRequests` to `100`
    c. (Debian, Ubuntu Precise, and Fedora) Update the prefork configuration 
       section to suit your environment. The following settings apply to a busy 
       system:
+
[source,bash]
------------------------------------------------------------------------------
<IfModule mpm_prefork_module>
   StartServers           20
   MinSpareServers         5
   MaxSpareServers        15
   MaxClients            150
   MaxRequestsPerChild 10000
</IfModule>
------------------------------------------------------------------------------
    d. (Ubuntu Trusty) As the *root* user, edit 
       /etc/apache2/mods-available/mpm_prefork.conf to match the above values.  
       Then, also as the *root* user, enable the mpm_prefork module by doing:
+
[source,bash]
------------------------------------------------------------------------------
a2dismod mpm_event
a2enmod mpm_prefork
------------------------------------------------------------------------------
+
6. (Fedora): As the *root* Linux account, edit the `/etc/httpd/eg_vhost.conf`
   file to change references from the non-existent `/etc/apache2/` directory
   to `/etc/httpd/`.
7. (Debian and Ubuntu Precise): As the *root* Linux account, enable the Evergreen site:
+
[source,bash]
------------------------------------------------------------------------------
a2dissite default  # OPTIONAL: disable the default site (the "It Works" page)
a2ensite eg.conf
------------------------------------------------------------------------------
(Ubuntu Trusty):
+
[source,bash]
------------------------------------------------------------------------------
a2dissite 000-default  # OPTIONAL: disable the default site (the "It Works" page)
a2ensite eg.conf
------------------------------------------------------------------------------

8. (Ubuntu): As the *root* Linux account, enable Apache to write
   to the lock directory; this is currently necessary because Apache
   is running as the `opensrf` user:
+
[source,bash]
------------------------------------------------------------------------------
chown opensrf /var/lock/apache2
------------------------------------------------------------------------------

Configure OpenSRF for the Evergreen application
-----------------------------------------------
There are a number of example OpenSRF configuration files in `/openils/conf/`
that you can use as a template for your Evergreen installation. Issue the
following commands as the *opensrf* Linux account:

[source, bash]
------------------------------------------------------------------------------
cp -b /openils/conf/opensrf_core.xml.example /openils/conf/opensrf_core.xml
cp -b /openils/conf/opensrf.xml.example /openils/conf/opensrf.xml
------------------------------------------------------------------------------

When you installed OpenSRF, you created four Jabber users on two
separate domains and edited the `opensrf_core.xml` file accordingly. Please
refer back to the OpenSRF README and, as the *opensrf* Linux account, edit the
Evergreen version of the `opensrf_core.xml` file using the same Jabber users
and domains as you used while installing and testing OpenSRF.

[NOTE]
The `-b` flag tells the `cp` command to create a backup version of the
destination file. The backup version of the destination file has a tilde (`~`)
appended to the file name, so if you have forgotten the Jabber users and
domains, you can retrieve the settings from the backup version of the files.

`eg_db_config`, described in the following section, sets the database
connection information in `opensrf.xml` for you.

Creating the Evergreen database
-------------------------------

Setting up the PostgreSQL server
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

For production use, most libraries install the PostgreSQL database server on a
dedicated machine. Therefore, by default, the `Makefile.install` prerequisite
installer does *not* install the PostgreSQL 9 database server that is required
by every Evergreen system. You can install the packages required by Debian or
Ubuntu Lucid on the machine of your choice using the following commands as the
*root* Linux account:

.(Debian / Ubuntu / Fedora) Installing PostgreSQL server packages

Each OS build target provides the postgres server installation packages
required for each operating system.  To install Postgres server packages, 
use the make target 'postgres-server-<OSTYPE>'.  Choose the most appropriate 
command below based on your operating system.

[source, bash]
------------------------------------------------------------------------------
make -f Open-ILS/src/extras/Makefile.install postgres-server-debian-wheezy
make -f Open-ILS/src/extras/Makefile.install postgres-server-debian-squeeze
make -f Open-ILS/src/extras/Makefile.install postgres-server-ubuntu-precise
make -f Open-ILS/src/extras/Makefile.install postgres-server-ubuntu-trusty
make -f Open-ILS/src/extras/Makefile.install postgres-server-fedora
------------------------------------------------------------------------------

.(Fedora) Postgres initialization

Installing Postgres on Fedora also requires you to initialize the PostgreSQL
cluster and start the service. Issue the following commands as the *root* user:

[source, bash]
------------------------------------------------------------------------------
postgresql-setup initdb
systemctl start postgresql
------------------------------------------------------------------------------

For a standalone PostgreSQL server, install the following Perl modules for your
distribution as the *root* Linux account:

.(Ubuntu Precise)
[source,bash]
---------------------------------------------------------------------------------
apt-get install gcc libxml-libxml-perl libxml-libxslt-perl libbusiness-isbn-perl
apt-get install libjson-xs-perl liblibrary-callnumber-lc-perl libmarc-record-perl
apt-get install libmarc-xml-perl libuuid-tiny-perl
cpan Rose::URI
---------------------------------------------------------------------------------

.(Debian "wheezy" and Ubuntu Trusty) 
standalone PostgreSQL 9 server
[source,bash]
---------------------------------------------------------------------------------
apt-get install gcc libxml-libxml-perl libxml-libxslt-perl libbusiness-isbn-perl
apt-get install libjson-xs-perl liblibrary-callnumber-lc-perl libmarc-record-perl
apt-get install libmarc-xml-perl libuuid-tiny-perl librose-uri-perl
---------------------------------------------------------------------------------

.(Fedora)
[source, bash]
------------------------------------------------------------------------------
yum install gcc perl-XML-LibXML perl-XML-LibXSLT perl-Business-ISBN
yum install perl-Library-CallNumber-LC perl-MARC-Record perl-MARC-Charset
yum install perl-MARC-File-XML perl-UUID-Tiny
------------------------------------------------------------------------------

You need to create a PostgreSQL superuser to create and access the database.
Issue the following command as the *postgres* Linux account to create a new
PostgreSQL superuser named `evergreen`. When prompted, enter the new user's
password:

[source, bash]
------------------------------------------------------------------------------
createuser -s -P evergreen
------------------------------------------------------------------------------

.Enabling connections to the PostgreSQL database

Your PostgreSQL database may be configured by default to prevent connections,
for example, it might reject attempts to connect via TCP/IP or from other
servers. To enable TCP/IP connections from localhost, check your `pg_hba.conf`
file, found in the `/etc/postgresql/` directory on Debian and Ubuntu, and in
the `/var/lib/pgsql/data/` directory on Fedora. A simple way to enable TCP/IP
connections from localhost to all databases with password authentication, which
would be suitable for a test install of Evergreen on a single server, is to
ensure the file contains the following entries _before_ any "host ... ident"
entries:

------------------------------------------------------------------------------
host    all             all             ::1/128                 md5
host    all             all             127.0.0.1/32            md5
------------------------------------------------------------------------------

When you change the `pg_hba.conf` file, you will need to reload PostgreSQL to
make the changes take effect.  For more information on configuring connectivity
to PostgreSQL, see
http://www.postgresql.org/docs/devel/static/auth-pg-hba-conf.html

Creating the Evergreen database and schema
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Once you have created the *evergreen* PostgreSQL account, you also need to
create the database and schema, and configure your configuration files to point
at the database server. Issue the following command as the *root* Linux account
from inside the Evergreen source directory, replacing <user>, <password>,
<hostname>, <port>, and <dbname> with the appropriate values for your
PostgreSQL database (where <user> and <password> are for the *evergreen*
PostgreSQL account you just created), and replace <admin-user> and <admin-pass>
with the values you want for the *egadmin* Evergreen administrator account:

[source, bash]
------------------------------------------------------------------------------
perl Open-ILS/src/support-scripts/eg_db_config --update-config \
       --service all --create-database --create-schema --create-offline \
       --user <user> --password <password> --hostname <hostname> --port <port> \
       --database <dbname> --admin-user <admin-user> --admin-pass <admin-pass>
------------------------------------------------------------------------------

This creates the database and schema and configures all of the services in
your `/openils/conf/opensrf.xml` configuration file to point to that database.
It also creates the configuration files required by the Evergreen `cgi-bin`
administration scripts, and sets the user name and password for the *egadmin*
Evergreen administrator account to your requested values.

You can get a complete set of options for `eg_db_config.pl` by passing the
`--help` parameter.

Loading sample data
~~~~~~~~~~~~~~~~~~~
If you add the `--load-all-sample` parameter to the `eg_db_config` command,
a set of authority and bibliographic records, call numbers, copies, staff
and regular users, and transactions will be loaded into your target
database. This sample dataset is commonly referred to as the _concerto_
sample data, and can be useful for testing out Evergreen functionality and
for creating problem reports that developers can easily recreate with their
own copy of the _concerto_ sample data.

Creating the database on a remote server
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
In a production instance of Evergreen, your PostgreSQL server should be
installed on a dedicated server.

PostgreSQL 9.1 and later
^^^^^^^^^^^^^^^^^^^^^^^^
To create the database instance on a remote database server running PostgreSQL
9.1 or later, simply use the `--create-database` flag on `eg_db_config`.

Starting Evergreen
------------------
1. As the *root* Linux account, start the `memcached` and `ejabberd` services
(if they aren't already running):
+
[source, bash]
------------------------------------------------------------------------------
/etc/init.d/ejabberd start
/etc/init.d/memcached start
------------------------------------------------------------------------------
+
2. As the *opensrf* Linux account, start Evergreen. The `-l` flag in the
following command is only necessary if you want to force Evergreen to treat the
hostname as `localhost`; if you configured `opensrf.xml` using the real
hostname of your machine as returned by `perl -ENet::Domain 'print
Net::Domain::hostfqdn() . "\n";'`, you should not use the `-l` flag.
+
[source, bash]
------------------------------------------------------------------------------
osrf_control -l --start-all
------------------------------------------------------------------------------
+
  ** If you receive the error message `bash: osrf_control: command not found`,
     then your environment variable `PATH` does not include the `/openils/bin`
     directory; this should have been set in the *opensrf* Linux account's
     `.bashrc` configuration file. To manually set the `PATH` variable, edit the
     configuration file `~/.bashrc` as the *opensrf* Linux account and add the
     following line:
+
[source, bash]
------------------------------------------------------------------------------
export PATH=$PATH:/openils/bin
------------------------------------------------------------------------------
+
3. As the *opensrf* Linux account, generate the Web files needed by the staff
   client and catalogue and update the organization unit proximity (you need to do
   this the first time you start Evergreen, and after that each time you change
   the library hierarchy in `config.cgi`):
+
[source, bash]
------------------------------------------------------------------------------
autogen.sh -u
------------------------------------------------------------------------------
+
4. As the *root* Linux account, restart the Apache Web server:
+
[source, bash]
------------------------------------------------------------------------------
/etc/init.d/apache2 restart
------------------------------------------------------------------------------
+
If the Apache Web server was running when you started the OpenSRF services, you
might not be able to successfully log in to the OPAC or staff client until the
Apache Web server is restarted.

Testing connections to Evergreen
--------------------------------

Once you have installed and started Evergreen, test your connection to
Evergreen via `srfsh`. As the *opensrf* Linux account, issue the following
commands to start `srfsh` and try to log onto the Evergreen server using the
*egadmin* Evergreen administrator user name and password that you set using the
`eg_db_config` command:

[source, bash]
------------------------------------------------------------------------------
/openils/bin/srfsh
srfsh% login <admin-user> <admin-pass>
------------------------------------------------------------------------------

You should see a result like:

    Received Data: "250bf1518c7527a03249858687714376"
    ------------------------------------
    Request Completed Successfully
    Request Time in seconds: 0.045286
    ------------------------------------

    Received Data: {
       "ilsevent":0,
       "textcode":"SUCCESS",
       "desc":" ",
       "pid":21616,
       "stacktrace":"oils_auth.c:304",
       "payload":{
          "authtoken":"e5f9827cc0f93b503a1cc66bee6bdd1a",
          "authtime":420
       }

    }

    ------------------------------------
    Request Completed Successfully
    Request Time in seconds: 1.336568
    ------------------------------------

If this does not work, it's time to do some troubleshooting.

  * As the *opensrf* Linux account, run the `settings-tester.pl` script to see
    if it finds any system configuration problems. The script is found at
    `Open-ILS/src/support-scripts/settings-tester.pl` in the Evergreen source
    tree.
  * Follow the steps in the http://evergreen-ils.org/dokuwiki/doku.php?id=troubleshooting:checking_for_errors[troubleshooting guide].
  * If you have faithfully followed the entire set of installation steps
    listed here, you are probably extremely close to a working system.
    Gather your configuration files and log files and contact the
    http://evergreen-ils.org/communicate/mailing-lists/[Evergreen development 
mailing list] for assistance before making any drastic changes to your system
    configuration.

Getting help
------------

Need help installing or using Evergreen? Join the mailing lists at
http://evergreen-ils.org/communicate/mailing-lists/ or contact us on the Freenode
IRC network on the #evergreen channel.

License
-------
This work is licensed under the Creative Commons Attribution-ShareAlike 3.0
Unported License. To view a copy of this license, visit
http://creativecommons.org/licenses/by-sa/3.0/ or send a letter to Creative
Commons, 444 Castro Street, Suite 900, Mountain View, California, 94041, USA.