Page tree

Unreleased Documentation

This documentation is unreleased and still in development. It may describe features which are not yet released in DSpace.
Looking for another version? See all documentation

Skip to end of metadata
Go to start of metadata

Installation Overview

Try out DSpace 7 before you install

If you'd like to quickly try out DSpace 7 before a full installation, see Try out DSpace 7 for instructions on a quick install via Docker.

As of version 7 (and above), the DSpace application is split into a "frontend" (User Interface) and a "backend" (Server API).  Most institutions will want to install BOTH.  However, you can decide whether to run them on the same machine or separate machines.

  • The DSpace Frontend consists of a User Interface built on Angular.io.  It cannot be run alone, as it requires a valid DSpace Backend to function.  The frontend provides all user-facing functionality
  • The DSpace Backend consists of a Server API ("server" webapp), built on Spring Boot.  It can be run standalone, however it has no user interface.  The backend provides all machine-based interfaces, including the REST API, OAI-PMH, SWORD (v1 and v2) and RDF.

We recommend installing the Backend first, as the Frontend requires a valid Backend to run properly.

Installing the Backend (Server API)

Work in progress (Feedback welcome)

These installation instructions are a work-in-progress and based heavily on the DSpace 6.x installation instructions.  Feedback or improvements are welcome.

Backend Requirements

UNIX-like OS or Microsoft Windows

  • UNIX-like operating system (Linux, HP/UX, Mac OSX, etc.) : Many distributions of Linux/Unix come with some of the dependencies below pre-installed or easily installed via updates.  You should consult your particular distribution's documentation or local system administrators to determine what is already available.
  • Microsoft Windows:  While DSpace can be run on Windows servers, most institutions tend to run it on a UNIX-like operating system.


Java JDK 11 (OpenJDK or Oracle JDK)

Make sure to install the JDK and not just the JRE

 At this time, DSpace requires the full JDK (Java Development Kit) be installed, rather than just the JRE (Java Runtime Environment).  So, please be sure that you are installing the full JDK and not just the JRE.


Only JDK11 is fully supported

Older versions of Java are unsupported. This includes JDK v7-10.

Newer versions of Java may work (e.g. JDK v12-16), but we do not recommend running them in Production.  We highly recommend running only Java LTS (Long Term Support) releases in Production, as non-LTS releases may not receive ongoing security fixes. As of this DSpace release, JDK11 is the most recent Java LTS release, with the next one (JDK17) being due sometime around September 2021.  As soon as the next Java LTS release is available, we will analyze it for compatibility with this release of DSpace.  For more information on Java releases, see the Java roadmaps for Oracle and/or OpenJDK.


Apache Maven 3.3.x or above (Java build tool)

Maven is necessary in the first stage of the build process to assemble the installation package for your DSpace instance. It gives you the flexibility to customize DSpace using the existing Maven projects found in the [dspace-source]/dspace/modules directory or by adding in your own Maven project to build the installation package for DSpace, and apply any custom interface "overlay" changes.

Maven can be downloaded from http://maven.apache.org/download.html

Configuring a Maven Proxy

You can configure a proxy to use for some or all of your HTTP requests in Maven. The username and password are only required if your proxy requires basic authentication (note that later releases may support storing your passwords in a secured keystore‚ in the meantime, please ensure your settings.xml file (usually ${user.home}/.m2/settings.xml) is secured with permissions appropriate for your operating system).

Example:

<settings>
  .
  .
  <proxies>
   <proxy>
      <active>true</active>
      <protocol>http</protocol>
      <host>proxy.somewhere.com</host>
      <port>8080</port>
      <username>proxyuser</username>
      <password>somepassword</password>
      <nonProxyHosts>www.google.com|*.somewhere.com</nonProxyHosts>
    </proxy>
  </proxies>
  .
  .
</settings>


Apache Ant 1.10.x or later (Java build tool)

While Apache Ant recommends using v1.10.x for Java 11, we've also had some success with recent versions of 1.9.x (specifically v1.9.15 seems to work fine with Java 11). That said, earlier versions of v1.9.x are not compatible with Java 11.

Apache Ant is required for the second stage of the build process (deploying/installing the application). First, Maven is used to construct the installer ([dspace-source]/dspace/target/dspace-installer), after which Ant is used to install/deploy DSpace to the installation directory.

Ant can be downloaded from the following location: http://ant.apache.org

Relational Database (PostgreSQL or Oracle)

PostgreSQL v11 (with pgcrypto installed)

PostgreSQL v9.4 to v11 will likely work, but earlier versions are less well tested.

Active development/testing on DSpace 7 has occurred on PostgreSQL v11.  However, it is likely that the backend would also function on PostgreSQL v9.4 - v10.  At this time we have not performed sufficient testing on these earlier versions to add them to the prerequisites listing.

DSpace 7 will definitely not function on versions below 9.4 as DSpace requires installing and running the pgcrypto extension (see below) v1.1, which was not available until PostgreSQL v9.4.

  • PostgreSQL can be downloaded from http://www.postgresql.org/.  It is also provided via many operating system package managers
  • Install the pgcrypto extension.  It will also need to be enabled on your DSpace Database (see Installation instructions below for more info). The pgcrypto extension allows DSpace to create UUIDs (universally unique identifiers) for all objects in DSpace, which means that (internal) object identifiers are now globally unique and no longer tied to database sequences.
    • On most Linux operating systems (Ubuntu, Debian, RedHat), this extension is provided in the "postgresql-contrib" package in your package manager. So, ensure you've installed "postgresql-contrib".
    • On Windows, this extension should be provided automatically by the installer (check your "[PostgreSQL]/share/extension" folder for files starting with "pgcrypto")
  • Unicode (specifically UTF-8) support must be enabled (but this is enabled by default).
  • Once installed, you need to enable TCP/IP connections (DSpace uses JDBC):
    • In postgresql.conf: uncomment the line starting: listen_addresses = 'localhost'.  This is the default, in recent PostgreSQL releases, but you should at least check it.
    • Then tighten up security a bit by editing pg_hba.conf and adding this line:

      host dspace dspace 127.0.0.1 255.255.255.255 md5

      This should appear before any lines matching all databases, because the first matching rule governs.

    • Then restart PostgreSQL.
Oracle 10g or later
  • Details on acquiring Oracle can be downloaded from the following location: http://www.oracle.com/database/. You will need to create a database for DSpace. Make sure that the character set is one of the Unicode character sets. DSpace uses UTF-8 natively, and it is suggested that the Oracle database use the same character set. You will also need to create a user account for DSpace (e.g. dspace) and ensure that it has permissions to add and remove tables in the database. Refer to the Quick Installation for more details.
    • NOTE: If the database server is not on the same machine as DSpace, you must install the Oracle client to the DSpace server and point tnsnames.ora and listener.ora files to the database the Oracle server.

Apache Solr 8.x or later (full-text index/search service)

Make sure to install Solr with Authentication disabled (which is the default).  DSpace does not yet support authentication to Solr (see https://github.com/DSpace/DSpace/issues/3169).  Instead, we recommend placing Solr behind a firewall and/or ensuring port 8983 (which Solr runs on) is not available for public/anonymous access on the web. Solr only needs to be accessible to requests from the DSpace backend.

Solr can be obtained at the Apache Software Foundation site for Lucene and Solr.  You may wish to read portions of the quick-start tutorial to make yourself familiar with Solr's layout and operation.  Unpack a Solr .tgz or .zip archive in a place where you keep software that is not handled by your operating system's package management tools, and arrange to have it running whenever DSpace is running.  You should ensure that Solr's index directories will have plenty of room to grow.  You should also ensure that port 8983 is not in use by something else, or configure Solr to use a different port.

If you are looking for a good place to put Solr, consider /opt or /usr/local.  You can simply unpack Solr in one place and use it.  Or you can configure Solr to keep its indexes elsewhere, if you need to – see the Solr documentation for how to do this.

It is not necessary to dedicate a Solr instance to DSpace, if you already have one and want to use it.  Simply copy DSpace's cores to a place where they will be discovered by Solr.  See below.

Servlet Engine (Apache Tomcat 9, Jetty, Caucho Resin or equivalent)

  • Apache Tomcat 9. Tomcat can be downloaded from the following location: http://tomcat.apache.org.  
    • The Tomcat owner (i.e. the user that Tomcat runs as) must have read/write access to the DSpace installation directory (i.e. [dspace])There are a few common ways this may be achieved:
      • One option is to specifically give the Tomcat user (often named "tomcat") ownership of the [dspace] directories, for example:

        # Change [dspace] and all subfolders to be owned by "tomcat"
        chown -R tomcat:tomcat [dspace]
      • Another option is to have Tomcat itself run as a new user named "dspace" (see installation instructions below).  Some operating systems make modifying the Tomcat "run as" user easily modifiable via an environment variable named TOMCAT_USER.  This option may be more desirable if you have multiple Tomcat instances running, and you do not want all of them to run under the same Tomcat owner.
    • You need to ensure that Tomcat has a) enough memory to run DSpace and b) uses UTF-8 as its default file encoding for international character support. So ensure in your startup scripts (etc) that the following environment variable is set: JAVA_OPTS="-Xmx512M -Xms64M -Dfile.encoding=UTF-8"
    • Modifications in [tomcat]/conf/server.xml : You also need to alter Tomcat's default configuration to support searching and browsing of multi-byte UTF-8 correctly. You need to add a configuration option to the <Connector> element in [tomcat]/config/server.xml: URIEncoding="UTF-8" e.g. if you're using the default Tomcat config, it should read:

      <!-- Define a non-SSL HTTP/1.1 Connector on port 8080 -->
      <Connector port="8080"
                    minSpareThreads="25"
                    enableLookups="false"
                    redirectPort="8443"
                    connectionTimeout="20000"
                    disableUploadTimeout="true"
                    URIEncoding="UTF-8"/>
      

      You may change the port from 8080 by editing it in the file above, and by setting the variable CONNECTOR_PORT in server.xml.  You should set the URIEncoding even if you are running Tomcat behind a proxy (Apache HTTPD, Nginx, etc.) via AJP.

  • Jetty or Caucho Resin 
    • DSpace 7 has not been tested with Jetty or Caucho Resin, after the switch to Java 11
    • Older versions of DSpace were able to run on a Tomcat-equivalent servlet Engine, such as Jetty (https://www.eclipse.org/jetty/) or Caucho Resin (http://www.caucho.com/). If you choose to use a different servlet container, please ensure that it supports Servlet Spec 3.1 (or above).
    • Jetty and Resin are configured for correct handling of UTF-8 by default.

(Optional) IP to City Database for Location-based Statistics

Optionally, if you wish to record the geographic locations of clients in DSpace usage statistics records, you will need to install (and regularly update) one of the following:

  • Either, a copy of MaxMind's GeoLite City database (in MMDB format)
    • NOTE: Installing MaxMind GeoLite2 is free.  However, you must sign up for a (free) MaxMind account in order to obtain a license key to use the GeoLite2 database.
    • You may download GeoLite2 directly from MaxMind, or many Linux distributions provide the geoipupdate tool directly via their package manager.  You will still need to configure your license key prior to usage.
    • Once the "GeoLite2-City.mmdb" database file is installed on your system,  you will need to configure its location as the value of usage-statistics.dbfile in your local.cfg configuration file
    • See the "Managing the City Database File" section of SOLR Statistics for more information about using a City Database with DSpace.
  • Or, you can alternatively use/install DB-IP's City Lite database (in MMDB format)
    • This database is also free to use, but does not require an account to download.
    • Once the "dbip-city-lite.mmdb" database file is installed on your system,  you will need to configure its location as the value of usage-statistics.dbfile in your local.cfg configuration file
    • See the "Managing the City Database File" section of SOLR Statistics for more information about using a City Database with DSpace.

Git (code version control)

Currently, there is a known bug in DSpace where a third-party Maven Module expects git to be available (in order to support the ./dspace version commandline tool).  We are working on a solution within this ticket:  DS-3418 - Getting issue details... STATUS

For the time being, you can work around this problem by installing Git locally: https://git-scm.com/downloads

Backend Installation

  1. Install all the Backend Requirements listed above.
  2. Create a DSpace operating system user (optional) .  As noted in the prerequisites above, Tomcat (or Jetty, etc) must run as an operating system user account that has full read/write access to the DSpace installation directory (i.e. [dspace]).  Either you must ensure the Tomcat owner also owns [dspace], OR you can create a new "dspace" user account, and ensure that Tomcat also runs as that account:

    useradd -m dspace
  3. Download the latest DSpace release from the DSpace GitHub Repository. You can choose to either download the zip or tar.gz file provided by GitHub, or you can use "git" to checkout the appropriate tag (e.g. dspace-7.0-beta5) or branch.
  4. Unpack the DSpace software. After downloading the software, based on the compression file format, choose one of the following methods to unpack your software:
    1. Zip file. If you downloaded dspace-7.0-beta5.zip do the following:

      unzip dspace-7.0-beta5.zip
    2. .gz file. If you downloaded dspace-7.0-beta.tar.gz do the following:

      gunzip -c dspace-7.0-beta5.tar.gz | tar -xf -

      For ease of reference, we will refer to the location of this unzipped version of the DSpace release as [dspace-source] in the remainder of these instructions. After unpacking the file, the user may wish to change the ownership of the dspace-7.x folder to the "dspace" user. (And you may need to change the group).

  5. Database Setup
    • PostgreSQL:
      • Create a dspace database user (this user can have any name, but we'll assume you name them "dspace"). This is entirely separate from the dspace operating-system user created above:

        createuser --username=postgres --no-superuser --pwprompt dspace

        You will be prompted (twice) for a password for the new dspace user.  Then you'll be prompted for the password of the PostgreSQL superuser (postgres).

      • Create a dspace database, owned by the dspace PostgreSQL user. Similar to the previous step, this can only be done by a "superuser" account in PostgreSQL (e.g. postgres):

        createdb --username=postgres --owner=dspace --encoding=UNICODE dspace

        You will be prompted for the password of the PostgreSQL superuser (postgres).

      • Finally, you MUST enable the pgcrypto extension on your new dspace database.  Again, this can only be enabled by a "superuser" account (e.g. postgres)

        # Login to the database as a superuser, and enable the pgcrypto extension on this database
        psql --username=postgres dspace -c "CREATE EXTENSION pgcrypto;"

        The "CREATE EXTENSION" command should return with no result if it succeeds. If it fails or throws an error, it is likely you are missing the required pgcrypto extension (see Database Prerequisites above).

        • Alternative method: How to enable pgcrypto via a separate database schema. While the above method of enabling pgcrypto is perfectly fine for the majority of users, there may be some scenarios where a database administrator would prefer to install extensions into a database schema that is separate from the DSpace tables. Developers also may wish to install pgcrypto into a separate schema if they plan to "clean" (recreate) their development database frequently. Keeping extensions in a separate schema from the DSpace tables will ensure developers would NOT have to continually re-enable the extension each time you run a "./dspace database clean". If you wish to install pgcrypto in a separate schema here's how to do that:

          # Login to the database as a superuser
          psql --username=postgres dspace
          # Create a new schema in this database named "extensions" (or whatever you want to name it)
          CREATE SCHEMA extensions;
          # Enable this extension in this new schema
          CREATE EXTENSION pgcrypto SCHEMA extensions;
          # Grant rights to call functions in the extensions schema to your dspace user
          GRANT USAGE ON SCHEMA extensions TO dspace;
          
          
          # Append "extensions" on the current session's "search_path" (if it doesn't already exist in search_path)
          # The "search_path" config is the list of schemas that Postgres will use
          SELECT set_config('search_path',current_setting('search_path') || ',extensions',false) WHERE current_setting('search_path') !~ '(^|,)extensions(,|$)';
          # Verify the current session's "search_path" and make sure it's correct
          SHOW search_path;
          # Now, update the "dspace" Database to use the same "search_path" (for all future sessions) as we've set for this current session (i.e. via set_config() above)
          ALTER DATABASE dspace SET search_path FROM CURRENT;
    • Oracle:
      • Setting up DSpace to use Oracle is a bit different now. You will need still need to get a copy of the Oracle JDBC driver, but instead of copying it into a lib directory you will need to install it into your local Maven repository. (You'll need to download it first from this location: http://www.oracle.com/technetwork/database/enterprise-edition/jdbc-112010-090769.html.) Run the following command (all on one line):

        mvn install:install-file
            -Dfile=ojdbc6.jar
            -DgroupId=com.oracle
            -DartifactId=ojdbc6
            -Dversion=11.2.0.4.0
            -Dpackaging=jar
            -DgeneratePom=true
        
      • You need to compile DSpace with an Oracle driver (ojdbc6.jar) corresponding to your Oracle version - update the version in [dspace-source]/pom.xml  E.g.:

        <dependency>
          <groupId>com.oracle</groupId>
          <artifactId>ojdbc6</artifactId>
          <version>11.2.0.4.0</version>
        </dependency>
        
      • Create a database for DSpace. Make sure that the character set is one of the Unicode character sets. DSpace uses UTF-8 natively, and it is required that the Oracle database use the same character set. Create a user account for DSpace (e.g. dspace) and ensure that it has permissions to add and remove tables in the database.
      • NOTE: You will need to ensure the proper db.* settings are specified in your local.cfg file (see next step), as the defaults for all of these settings assuming a PostgreSQL database backend.

        db.url = jdbc:oracle:thin:@host:port/SID
        # e.g. db.url = jdbc:oracle:thin:@//localhost:1521/xe
        # NOTE: in db.url, SID is the SID of your database defined in tnsnames.ora
        # the default Oracle port is 1521
        # You may also use a full SID definition, e.g.
        # db.url = jdbc:oracle:thin:@(description=(address_list=(address=(protocol=TCP)(host=localhost)(port=1521)))(connect_data=(service_name=DSPACE)))
        
        # Oracle driver and dialect
        db.driver = oracle.jdbc.OracleDriver
        db.dialect = org.hibernate.dialect.Oracle10gDialect
        
        # Specify DB username, password and schema to use
        db.username =
        db.password =
        db.schema = ${db.username}
        # For Oracle, schema is equivalent to the username of your database account,
        # so this may be set to ${db.username} in most scenarios
      • Later, during the Maven build step, don't forget to specify mvn -Ddb.name=oracle package

  6. Initial Configuration (local.cfg):  Create your own [dspace-source]/dspace/config/local.cfg configuration file (you may wish to simply copy the provided [dspace-source]/dspace/config/local.cfg.EXAMPLE). This local.cfg file can be used to store any configuration changes that you wish to make which are local to your installation (see local.cfg configuration file documentation). ANY setting may be copied into this local.cfg file from the dspace.cfg or any other *.cfg file in order to override the default setting (see note below).  For the initial installation of DSpace, there are some key settings you'll likely want to override, those are provided in the [dspace-source]/dspace/config/local.cfg.EXAMPLE. (NOTE: Settings followed with an asterisk (*) are highly recommended, while all others are optional during initial installation and may be customized at a later time)
    • dspace.dir* - must be set to the [dspace] (installation) directory  (NOTE: On Windows be sure to use forward slashes for the directory path!  For example: "C:/dspace" is a valid path for Windows.)
    • dspace.server.url* - complete URL of this DSpace backend (including port and any subpath).  For example: http://localhost:8080/server/
    • dspace.ui.url* - complete URL of the DSpace frontend (including port and any subpath). REQUIRED for the REST API to fully trust requests from the DSpace frontend. For example: http://localhost:4000/
    • dspace.name - "Proper" name of your server, e.g. "My Digital Library".
    • solr.server* - complete URL of the Solr server. DSpace makes use of Solr for indexing purposes.  http://localhost:8983/solr unless you changed the port or installed Solr on some other host.
    • default.language - Default language for all metadata values (defaults to "en_US")
    • db.url* - The full JDBC URL to your database (examples are provided in the local.cfg.EXAMPLE)
    • db.driver* - Which database driver to use, based on whether you are using PostgreSQL or Oracle
    • db.dialect* - Which database dialect to use, based on whether you are using PostgreSQL or Oracle
    • db.username* - the database username used in the previous step.
    • db.password* - the database password used in the previous step.
    • db.schema* - the database scheme to use (examples are provided in the local.cfg.EXAMPLE)
    • mail.server - fully-qualified domain name of your outgoing mail server.
    • mail.from.address - the "From:" address to put on email sent by DSpace.
    • mail.feedback.recipient - mailbox for feedback mail.
    • mail.admin - mailbox for DSpace site administrator.
    • alert.recipient - mailbox for server errors/alerts (not essential but very useful!)
    • registration.notify- mailbox for emails when new users register (optional)

      Your local.cfg file can override ANY settings from other *.cfg files in DSpace

      The provided local.cfg.EXAMPLE only includes a small subset of the configuration settings available with DSpace. It provides a good starting point for your own local.cfg file.

      However, you should be aware that ANY configuration can now be copied into your local.cfg to override the default settings.  This includes ANY of the settings/configurations in:

      • The primary dspace.cfg file ([dspace]/config/dspace.cfg)
      • Any of the module configuration files ([dspace]/config/modules/*.cfg files)
      • Any of the Spring Boot settings ([dspace-src]/dspace-server-webapp/src/main/resources/application.properties)

      Individual settings may also be commented out or removed in your local.cfg, in order to re-enable default settings.

      See the Configuration Reference section for more details.

  7. DSpace Directory: Create the directory for the DSpace backend installation (i.e. [dspace]). As root (or a user with appropriate permissions), run:

    mkdir [dspace]
    chown dspace [dspace]

    (Assuming the dspace UNIX username.)

  8. Build the Installation Package: As the dspace UNIX user, generate the DSpace installation package.

    cd [dspace-source]
    mvn package
    

    Building with Oracle Database Support

    Without any extra arguments, the DSpace installation package is initialized for PostgreSQL. If you want to use Oracle instead, you should build the DSpace installation package as follows:
    mvn -Ddb.name=oracle package

  9. Install DSpace: As the dspace UNIX user, install DSpace to [dspace]:

    cd [dspace-source]/dspace/target/dspace-installer
    ant fresh_install

    To see a complete list of build targets, run: ant help The most likely thing to go wrong here is the test of your database connection. See the Installing DSpace (OLD - to be removed)#Common Problems Section below for more details.

  10. Deploy Server web application: The DSpace backend consists of a single "server" webapp (in [dspace]/webapps/server).  You need to deploy this webapp into your Servlet Container (e.g. Tomcat).  Generally, there are two options (or techniques) which you could use...either configure Tomcat to find the DSpace "server" webapp, or copy the "server" webapp into Tomcat's own webapps folder.
    • Technique A. Tell your Tomcat/Jetty/Resin installation where to find your DSpace web application(s). As an example, in the directory [tomcat]/conf/Catalina/localhost you could add files similar to the following (but replace [dspace]with your installation location):

      DEFINE A CONTEXT PATH FOR DSpace Server webapp: server.xml
      <?xml version='1.0'?>
      <Context
      	docBase="[dspace]/webapps/server"/>

      The name of the file (not including the suffix ".xml") will be the name of the context, so for example server.xml defines the context at http://host:8080/server.  To define the root context (http://host:8080/), name that context's file ROOT.xml.   Optionally, you can also choose to install the old, deprecated "rest" webapp if you

    • Technique B. Simple and complete. You copy only (or all) of the DSpace Web application(s) you wish to use from the [dspace]/webapps directory to the appropriate directory in your Tomcat/Jetty/Resin installation. For example:
      cp -R [dspace]/webapps/* [tomcat]/webapps* (This will copy all the web applications to Tomcat).
      cp -R [dspace]/webapps/server [tomcat]/webapps* (This will copy only the Server web application to Tomcat.)

      To define the root context (http://host:8080/), name that context's directory ROOT.

  11. Optionally, also install the deprecated DSpace 6.x REST API web application.  If you previously used the DSpace 6.x REST API, for backwards compatibility the old, deprecated "rest" webapp is still available to install (in [dspace]/webapps/rest). It is NOT used by the DSpace frontend.  So, most users should skip this step.
  12. Copy Solr cores:  DSpace installation creates a set of four empty Solr cores already configured.  Copy them from [dspace]/solr to the place where your Solr instance will discover them.  Start (or re-start) Solr.  For example:

    cp -R [dspace]/solr/* [solr]/server/solr/configsets
    [solr]/bin/solr restart


    You can check the status of Solr and your new DSpace cores by using its administrative web interface.  Browse to http://localhost:8983/ to see if Solr is running well, then look at the cores by selecting (on the left) Core Admin or using the Core Selector drop list.

  13. Create an Administrator Account:  Create an initial administrator account from the command line:

    [dspace]/bin/dspace create-administrator


  14. Initial Startup!  Now the moment of truth! Start up (or restart) Tomcat/Jetty/Resin.
    1. REST API Interface - (e.g.)  http://dspace.myu.edu:8080/server/
    2. OAI-PMH Interface - (e.g.)  http://dspace.myu.edu:8080/server/oai/request?verb=Identify
    3. For an example of what the default backend looks like, visit the Demo Backend: https://api7.dspace.org/server/
  15. Production Installation (adding HTTPS support): Running the DSpace Backend on HTTP & port 8080 is only usable for testing/demo environments.  If you want to run DSpace in Production, you MUST run the backend with HTTPS support (otherwise logins will not work outside of your local domain).
    1. For HTTPS support, we recommend installing either Apache HTTPD or Nginx, configuring SSL at that level, and proxying all requests to your Tomcat installation (on port 8080).  Keep in mind, if you want to host both the DSpace Backend and Frontend on the same server, you can use one installation of Apache HTTPD or Nginx to manage HTTPS/SSL and proxy to both.
    2. These instructions are specific to Apache HTTPD, but a similar setup can be achieved with Nginx
      1. Install Apache HTTPD, e.g. sudo apt install apache2
      2. Install the mod_proxy and mod_proxy_ajp modules, e.g. sudo en2mod proxy; sudo a2enmod proxy_ajp
        1. Alternatively, you can choose to use mod_proxy_http to create an http proxy.  A separate example is commented out below

      3. Restart Apache to enable
      4. For mod_proxy_ajp to communicate with Tomcat, you'll need to enable Tomcat's AJP connector in your Tomcat's server.xml:

        <Connector protocol="AJP/1.3" port="8009" redirectPort="8443" URIEncoding="UTF-8" />
      5. Now, setup a new VirtualHost for your site (using HTTPS / port 443) which proxies all requests to Tomcat's AJP connector (running on port 8009)

        <VirtualHost _default_:443>
            .. setup your host how you want, including log settings...
        
            SSLEngine on
            SSLCertificateFile [full-path-to-PEM-cert]
            SSLCertificateKeyFile [full-path-to-cert-KEY]
        
            # Proxy all HTTPS requests to "/server" from Apache to Tomcat via AJP connector
            ProxyPass /server ajp://localhost:8009/server
            ProxyPassReverse /server ajp://localhost:80009/server
        
            # If you would rather use mod_proxy_http as an http proxy to port 8080
            # then use these settings instead
            #ProxyPass /server http://localhost:8080/server
            #ProxyPassReverse /server http://localhost:8080/server
            # When using mod_proxy_http, you need to also ensure the X-Forwarded-Proto header is sent
            # to tell DSpace it is behind HTTPS, otherwise some URLs may continue to use HTTP 
            $ (requires installing/enabling mod_headers)
            #RequestHeader set X-Forwarded-Proto https
        </VirtualHost>
    3. After switching to HTTPS, make sure to go back and update the URLs (e.g. dspace.server.url) in your local.cfg to match the new URL of your backend.  This will require briefly rebooting Tomcat.

Installing the Frontend (User Interface)

Work in progress (Feedback welcome)

These installation instructions are a work-in-progress.  They do NOT yet include production Link -ready installation scenarios for running the (Angular) frontend via production tools like PM2 or Passenger.  Feedback or improvements are welcome.

Frontend Requirements

UNIX-like OS or Microsoft Windows

  • UNIX-like operating system (Linux, HP/UX, Mac OSX, etc.) : Many distributions of Linux/Unix come with some of the dependencies below pre-installed or easily installed via updates.  You should consult your particular distribution's documentation or local system administrators to determine what is already available.
  • Microsoft Windows:  While DSpace can be run on Windows servers, most institutions tend to run it on a UNIX-like operating system.

Node.js (v12.x or v14.x)

  • Node.js can be found at https://nodejs.org/.  It may be available through your Linux distribution's package manager.  We recommend running a Long Term Support (LTS) version (even numbered releases).  Non-LTS versions (odd numbered releases) are not recommended.
  • Node.js is a Javascript runtime that also provides npm (Node Package Manager). It is used to both build and run the frontend.

Yarn (v1.x)

  • Yarn v1.x is available at https://classic.yarnpkg.com/.  It can usually be install via NPM (or through your Linux distribution's package manager)

    # You may need to run this command using "sudo" if you don't have proper privileges
    npm install --global yarn
  • Yarn is used to build/install the frontend.

PM2 (or another Process Manager for Node.js apps) (optional, but recommended for Production)

  • In Production scenarios, we highly recommend starting/stopping the User Interface using a Node.js process manager. There are several available, but our current favorite is PM2.  The rest of this installation guide assumes you are using PM2.
  • PM2 is very easily installed via NPM

    # You may need to run this command using "sudo" if you don't have proper privileges
    npm install --global pm2

DSpace 7.x Backend (see above)

  • The DSpace User Interface (Frontend) cannot function without an installed DSpace Backend.  Follow the instructions above.
  • The Frontend and Backend do not need to be installed on the same machine/server.  They may be installed on separate machines as long as the two machines can connect to one another via HTTP or HTTPS.

Frontend Installation

  1. First, install all the Frontend Requirements listed above & verify the backend/REST API is publicly accessible.
  2. Download the latest dspace-angular release from the DSpace GitHub repository. You can choose to either download the zip or tar.gz file provided by GitHub, or you can use "git" to checkout the appropriate tag (e.g. dspace-7.0-beta5) or branch.
  3. Install all necessary local dependencies by running the following from within the unzipped "dspace-angular" directory

    # change directory to our repo
    cd dspace-angular
    
    # install the local dependencies
    yarn install
  4. Create a Production Configuration file at [dspace-angular]/src/environments/environment.prod.ts. You may wish to use the environment.template.ts as a starting point. This environment.prod.ts file can be used to override any of the default configurations specified in the environment.common.ts (in that same directory).  At a minimum this file MUST include the "ui" and "rest" sections similar to the following (keep in mind, you only need to include settings that you need to modify): 

    export const environment = {   
      // The "ui" section defines where you want Node.js to run/respond. It may correspond to your primary URL, but it also may not (if you are running behind a proxy).
      // In this example, we are setting up our UI to just use localhost, port 4000. 
      // This is a common setup for when you want to use Apache or Nginx to handle HTTPS and proxy requests to Node on port 4000
      ui: { 
          ssl: false,
          host: 'localhost',
          port: 4000, 
          // NOTE: Space is capitalized because 'namespace' is a reserved string in TypeScript
          nameSpace: '/' 
      },
      // This example is valid if your Backend is publicly available at https://api.mydspace.edu/server/
      // The REST settings MUST correspond to the primary URL of the backend. Usually, this means they must be kept in sync
      // with the value of "dspace.server.url" in the backend's local.cfg
      rest: { 
          ssl: true, 
          host: 'api.mydspace.edu',
          port: 443, 
          // NOTE: Space is capitalized because 'namespace' is a reserved string in TypeScript
          nameSpace: '/server' 
      }
    };
    1. HINT #1: In the "ui" section above, you may wish to start with "ssl: false" and "port: 4000" just to be certain that everything else is working properly.  With those settings, you can quickly test your UI by running "yarn start" and trying to access it via http://[mydspace.edu]:4000/ from your web browser.  KEEP IN MIND, we highly recommend always using HTTPS for Production.
    2. HINT #2: If Node throws an error saying "listen EADDRNOTAVAIL: address not available", try setting the "host" to "0.0.0.0" or "localhost".  Usually that error is a sign that the "host" is not recognized.
    3. If there are other settings you know you need to modify in the default environment.common.ts configuration file you can also copy them into this same file.
  5. Build the User Interface for Production. This uses your environment.prod.ts and the source code to create a compiled version of the UI in the [dspace-angular]/dist folder

    yarn run build:prod
    1. HINT: if you change/update your environment.prod.ts, then you will need to rebuild the UI application (i.e. rerun this command).
  6. Assuming you are using PM2, create a JSON configuration file describing how to run our UI application.  This need NOT be in the same directory as the dspace-angular codebase itself (in fact you may want to put the parent directory or another location). Keep in mind the "cwd" setting (on line 5) must be the full path to your [dspace-angular] folder.

    dspace-angular.json
    {
        "apps": [
            {
                "name": "dspace-angular",
                "cwd": "/home/dspace/dspace-angular",
                "script": "yarn",
                "args": "run serve:ssr",
               "interpreter": "none"
            }
        ]
    }
  7. Now, start the application using PM2 using the configuration file you created in the previous step

    # In this example, we are assuming the config is named "dspace-angular.json"
    pm2 start dspace-angular.json
    
    # To see the logs, you'd run
    # pm2 logs
    
    # To stop it, you'd run
    # pm2 stop dspace-angular.json
    1. For more PM2 commands see https://pm2.keymetrics.io/docs/usage/quick-start/
    2. HINT: You may also want to install/configure pm2-logrotate to ensure that PM2's log folder doesn't fill up over time.
  8. At this point, the User Interface should be available at the URL you configured in your environment.prod.ts
    1. For an example of what the default frontend looks like, visit the Demo Frontend: https://demo7.dspace.org/ 
  9. For HTTPS (port 443) support, you have two options
    1. (Recommended) You can install either Apache HTTPD or Nginx , configuring SSL at that level, and proxy requests to PM2 (on port 4000).  This is our current recommended approach. Plus, as a bonus, if you want to host the UI and Backend on the same server, you can use just one Apache HTTPD (or Nginx) to proxy to both. These instructions are specific to Apache.
      1. Install Apache HTTPD, e.g. sudo apt install apache2
      2. Install the mod_proxy and mod_proxy_http modules, e.g. sudo en2mod proxy; sudo a2enmod proxy_http
      3. Restart Apache to enable
      4. Now, setup a new VirtualHost for your site (preferably using HTTPS / port 443) which proxies all requests to PM2 running on port 4000.

        <VirtualHost _default_:443>
            .. setup your host how you want, including log settings...
        
            SSLEngine on
            SSLCertificateFile [full-path-to-PEM-cert]
            SSLCertificateKeyFile [full-path-to-cert-KEY]
        
            # Proxy all HTTPS requests from Apache to PM2 on port 4000
            ProxyPass / http://127.0.0.1:4000/
            ProxyPassReverse / http://127.0.0.1:4000/
        </VirtualHost>
    2. (Alternatively) You can use the basic HTTPS support built into dspace-angular node server.  (This may currently be better for non-Production environments as it has not been well tested) 
      1. Create a [dspace-angular]/config/ssl/ folder and add a key.pem and cert.pem to that folder (they must have those exact names)
      2. Enable "ui.ssl" (set to true)
      3. Update your "ui.port" to be 443
        1. In order to run Node/PM2 on port 443, you also will likely need to provide node with special permissions, like in this example.
      4. Rebuild and then restart the app in PM2
      5. Keep in mind, while this setup is simple, you may not have the same level of detailed, Production logs as you would with Apache HTTPD or Nginx
  10. Additional UI configurations are described in the environment.common.ts and at https://github.com/DSpace/dspace-angular/blob/main/docs/Configuration.md  (More documentation will be coming soon)

What Next?

After a successful installation, you may want to take a closer look at

If you've run into installation problems, you may want to...

  • Review commons installation issues (LINK COMING)
  • Ask for Support via one of the support options documented on that page

Common Installation Issues

"CORS error" or "Invalid CORS request"

If you are seeing a CORS error in your browser, this means that you are accessing the REST API via an "untrusted" client application.  To fix this error, you must change your REST API / Backend configuration to trust the application.

  • By default, the DSpace REST API / Backend will only trust the application at dspace.ui.url.  Therefore, you should first verify that your dspace.ui.url setting (in your local.cfg) exactly matches the primary URL of your User Interface (i.e. the URL you see in the browser).  This must be an exact match: mode (http vs https), domain, port, and subpath(s) all must match.
  • If you need to trust additional client applications / URLs, those MUST be added to the rest.cors.allowed-origins configuration. See REST API for details on this configuration.
  • Also, check your Tomcat (or servlet container) log files. If Tomcat throws a syntax or other major error, it may return an error response that triggers a CORS error.  In this scenario, the CORS error is only a side effect of a larger error.

If you modify either of the above settings, you will need to restart Tomcat for the changes to take effect.

"403 Forbidden" error with a message that says "Access is denied. Invalid CSRF Token"

First, double check that you are seeing that exact error message.  A 403 Forbidden error may be thrown in a variety of scenarios. For example, a 403 may be thrown if a page requires a login, if you have entered an invalid username or password, or even sometimes when there is a CORS error (see previous installation issue for how to solve that). 

If you are seeing the message "Invalid CSRF Token" message (especially on every login), this is usually the result of a configuration / setup issue.

Here's some things you should double check:

  • If you need to be able to login to the REST API from other domains, then your Backend must be running HTTPS.  
    • If the REST API Backend is running HTTP, then it will always send the required DSPACE-XSRF-COOKIE cookie with a value of SameSite=Lax.  This setting means that the cookie will not be sent (by your browser) to any other domains. Effectively, this will block all logins from any domain that is not the same as the REST API (as this cookie will not be sent back to the REST API as required for CSRF validation).  In other words, running the REST API on HTTP is only possible if the User Interface is running on the exact same domain. For example, running both on 'localhost' with HTTP is a common development setup, and this will work fine.
    • In order to allow for cross-domain logins, you MUST enable HTTPS on the REST API. This will result in the DSPACE-XSRF-COOKIE cookie being set to SameSite=None; Secure.  This setting means the cookie will be sent cross domain, but only for HTTPS requests. It also allows the user interface (or other client applications) to be on any domain, provided that the domain is trusted by CORS (see rest.cors.allowed-origins setting in REST API)
  • Verify that your User Interface's "rest" section matches the value of "dspace.server.url" configuration on the Backend.  This simply ensures your UI is sending requests to the correct REST API.  Also pay close attention that both specify HTTPS when necessary (see previous bullet).
  • Verify that your "dspace.server.url" configuration on the Backend matches the primary URL of the REST API (i.e. the URL you see in the browser).  This must be an exact match: mode (http vs https), domain, port, and subpath(s) all must match.
  • Verify that your "dspace.ui.url" configuration on the Backend matches the primary URL of your User Interface (i.e. the URL you see in the browser). This must be an exact match: mode (http vs https), domain, port, and subpath(s) all must match.
  • If you are running a custom application, or accessing the REST API from the command-line (or other third party tool like Postman), you MUST ensure you are sending the CSRF token on every modifying request.  See our REST Contract for more details https://github.com/DSpace/RestContract/blob/main/csrf-tokens.md

For additional information on how DSpace's CSRF Protection works, see our REST Contract at https://github.com/DSpace/RestContract/blob/main/csrf-tokens.md

Using a Self-Signed SSL Certificate causes the Frontend to not be able to access the Backend

If you setup the backend to use HTTPS with a self-signed SSL certificate, then Node.js (which the frontend runs on) may not "trust" that certificate by default.  This will result in the Frontend not being able to make requests to the Backend.

One possible workaround (untested as of yet) is to try setting the NODE_EXTRA_CA_CERTS environment variable (which tells Node.js to trust additional CA certificates).

Another option is to avoid using a self-signed SSL certificate. Instead, create a real, issued SSL certificate using something like Let's Encrypt (or similar free services)

My REST API is running under HTTPS, but some of its "link" URLs are switching to HTTP?

This scenario may occur when you are running the REST API behind an HTTP proxy (e.g. Apache HTTPD's mod_proxy_http, Ngnix's proxy_pass or any other proxy that is forwarding from HTTPS to HTTP).

The fix is to ensure the DSpace REST API is sent the X-Forwarded-Proto header (by your proxying service), telling it that the forwarded protocol is HTTPS

X-Forwarded-Proto: https

In general, when running behind a proxy, the DSpace REST API depends on accurate X-Forwarded-* headers to be sent by that proxy.


  • No labels

1 Comment

  1. I was trying to install the Backend (Server API) locally. Everything went smooth except that none of the database tables were created. This lead to the server application not being properly deployed by tomcat. Running: "./bin/dspace database migrate" after step 9 fixed it.