The quick-start guide showed how to quickly launch a temporary instance of Apollo, but deploying the application to production normally involves some extra steps.
The general idea behind your deployment is to create a
apollo-config.groovy file from some existing sample files which
have sample settings for various database engines.
You will minimally need to have Java 8 or greater, Grails, git, ant, a servlet container e.g. tomcat7+, jetty, or resin. An external database such as PostgreSQL or MySQL is generally used for production, but instructions for the H2 database is also provided.
Important note: The default memory for Tomcat and Jetty is insufficient to run Apollo (and most other web apps).You should increase the memory according to these instructions.
Other possible build settings for JBrowse (based on an Ubuntu 16 install):
sudo apt-get update && sudo apt-get install zlib1g-dev libpng-dev libgd2-noxpm-dev build-essential git python-software-properties curl -sL https://deb.nodesource.com/setup_6.x | sudo -E bash - sudo apt-get install nodejs
NOTE: npm (installed with nodejs) must be version 2 or better. If not installed from the above instructions, most stable versions of node.js will supply this.
sudo npm install -g bower
NOTE: you must link nodejs to to node if your system installs it as a
nodejs binary instead of a node one. E.g.,
sudo ln -s /usr/bin/nodejs /usr/bin/node
Build settings for Apollo specifically. Recent versions of tomcat7 will work, though tomcat8 is preferred. If it does not install automatically there are a number of ways to build tomcat on linux:
sudo apt-get install ant openjdk-8-jdk export JAVA_HOME=/usr/lib/jvm/java-8-openjdk-amd64/ # or set in .bashrc / .project
Download Apollo from the latest release under source-code and unzip. Test installation by running
./apollo run-local and see that the web-server starts up on http://localhost:8080/apollo/. To setup for production continue onto configuration below after install .
If you get an
Unsupported major.minor error or similar, please confirm that the version of java that tomcat is running
ps -ef | grep java is the same as the one you used to build. Setting JAVA_HOME to the Java 8 JDK should fix most problems.
Apollo supports several database backends, and you can choose sample configurations from using H2, Postgres, or MySQL by default.
Each has a file called
sample-postgres-apollo-config.groovy that is designed to be
renamed to apollo-config.groovy before running
apollo deploy. Additionally there is a
sample-docker-apollo-config.groovy which allows control of the configuration via environment variables.
apollo-config.groovy has different groovy environments for test, development, and production modes.
The environment will be selected automatically selected depending on how it is run, e.g:
apollo releaseuse the production environment (i.e. when you copy the war file to your production server
apollo debuguse the development environment (i.e. when you are running it locally)
apollo testuses the test environment (i.e. only when running unit tests)
Configure for H2:¶
- H2 is an embedded database engine, so no external setups are needed. Simply copy sample-h2-apollo-config.groovy to
- The default dev environment (
apollo run-app) is in memory so you will have to change that to file.
- The default dev environment (
- If you use H2 with tomcat or jetty in production you have to set the permissions for the file path in production correctly (e.g.
chown -u tomcat:tomcat /mypath/prodDb.*.db).
- If you use the local relative path
jdbc:h2:./prodDband tomcat8 the path will likely be:
- If you use the local relative path
Configure for PostgreSQL:¶
- Create a new database with postgres and add a user for production mode. Here are a few ways to do this in PostgreSQL.
- Copy the sample-postgres-apollo-config.groovy to apollo-config.groovy.
Configure for MySQL:¶
- Create a new MySQL database for production mode (i.e. run ``create database `apollo-production``` in the mysql console) and copy the sample-postgres-apollo-config.groovy to apollo-config.groovy.
Configure for Docker:¶
- Set up and export all of the environment variables you wish to configure. At bare minimum you will likely wish to set
- Create a new database in your chosen database backend and copy the sample-docker-apollo-config.groovy to apollo-config.groovy.
- Instructions and a script for launching docker with apollo and PostgreSQL.
After you startup the application, the database schema (tables, etc.) is automatically setup. You don’t have to initialize any database schemas yourself.
Deploy the application¶
apollo run-local command only launches a temporary server and should really not be used in production, so to
deploy to production, we build a new WAR file with the
apollo deploy command. After you have setup your
apollo-config.groovy file, and it has the appropriate username, password, and JDBC URL in it, then we can run the
This command will package the application and it will download any missing pre-requisites (jbrowse) into a WAR file in
the “target/” subfolder. After it completes, you can then copy the WAR file (e.g.
apollo-2.0.4.war) from the target folder
web-app folder of your web container installation.
If you name the file
apollo.war in your webapps folder, then you can access your app at “http://localhost:8080/apollo”
Alternatively, as we alluded to previously, you can also launch a temporary instance of the server which is useful for testing
./apollo run-local 8085
This temporary server will be accessible at “http://localhost:8085/apollo”
If you have tracks that have deep nested features that will result in a feature JSON larger than 10MB or if you have a client
that sends requests to the Apollo server as JSON of size larger than 10MB then you will have to modify
Specifically the following block in
<context-param> <param-name>org.apache.tomcat.websocket.textBufferSize</param-name> <param-value>10000000</param-value> </context-param> <context-param> <param-name>org.apache.tomcat.websocket.binaryBufferSize</param-name> <param-value>10000000</param-value> </context-param>
<param-value> is in bytes.
Changing the memory used by Apollo in production must be configured within Tomcat directly.
The default memory assigned to Apollo to run commands in Apollo is 2048 MB. This can be changed in your
apollo-config.groovy by uncommenting the memory configuration block:
// Uncomment to change the default memory configurations grails.project.fork = [ test : false, // configure settings for the run-app JVM run : [maxMemory: 2048, minMemory: 64, debug: false, maxPerm: 1024, forkReserve: false], // configure settings for the run-war JVM war : [maxMemory: 2048, minMemory: 64, debug: false, maxPerm: 1024, forkReserve: false], // configure settings for the Console UI JVM console: [maxMemory: 2048, minMemory: 64, debug: false, maxPerm: 1024] ]
Note on database settings¶
If you use the
apollo run-local command, then the “development” section of the apollo-config.groovy is used (or an
temporary in-memory H2 database is used if no apollo-config.groovy exists).
If you use the WAR file generated by the
apollo deploy command on your own webserver, then the “production” section of
the apollo-config.groovy is used.
Detailed build instructions¶
While the shortcut
apollo deploy takes care of basic application deployment, understanding the full build process of
Apollo can help you to optimize and improve your deployed instances.
To learn more about the architecture of webapollo, view the architecture guide but the main idea here
is to learn how to use
sudo apt-get install nodejs sudo yum install epel-release npm brew install node
Install extra perl modules¶
Building apollo in release mode also requires some extra Perl modules, namely Text::Markdown and DateTime. One way to install them:
bin/cpanm -l extlib DateTime Text::Markdown
In all other respects,
apollo release is exactly the same as
apollo deploy though.
Performing active development¶
To perform active development of the codebase, use
This will launch a temporary instance of Apollo by running
grails run-app and
ant devmode at the same time,
which means that any changes to the Java files will be picked up, allowing fast iteration.
scripts/copy_client.sh and these will be
picked up on-the-fly too.