Skip to content

Diabetes portal for the Medical Population Genetics department at the Broad Institute

Notifications You must be signed in to change notification settings

broadinstitute/dig-diabetes-portal

Repository files navigation

__Important note to portal developers and sysops __ __ The code requires the incorporation of a private configuration file, as described below in the section titled "Configuration". The code may compile correctly without private configuration files, but it will not run in any of the following configurations:

  • debug mode for development
  • test mode, for running unit, integration, or functional tests
  • generating a war file for the purposes of deployment
  • running that war file on your target server More information is provided down below (see the "configuration" section in this README file)__

Diabetes portal

The diabetes portal is written in Grails. The following description should give you all the information you need to download, compile, and run a local version of the portal.

Table of Contents:

Environment
Set up your environment for running a Grails based project
Get the code
Download the code from GitHub
Testing
Run some pre-packaged tests to make sure your environment is in good shape
Run the portal
Start up the diabetes portal and run it locally

Environment

The following steps should give you a working development environment:

Install Java

Googling “download Java jdk” should take you where you need to go. The same command brought me to this URL from which I successfully downloaded Java.

Note that you will need the whole Java development toolkit (JDK), not simply the Java runtime environment (JRE). As far as versions, I have successfully compiled this program both under Java 6 and Java 7 (I haven't yet tried Java 8).

Once you’ve downloaded Java you should run the installer. Make sure that one way or another you end up with an environmental variable named “JAVA_HOME” which points at the base directory of your JDK. I usually perform this assignment inside my .bash_profile with a few lines that look like this:

JAVA_HOME=/cygdrive/c/java7
PATH=$JAVA_HOME/bin:$PATH
export JAVA_HOME

Other approaches to defining an environmental variable are fine too. Simply make sure that if you type:

echo $JAVA_HOME

that you see the directory you expect to see. The critical test of your success in this step, however, is typing

java -version

which should report the version of your Java runtime environment.

Install GVM

There’s a tool called GVM that can do most of the rest of the installation work for you. You can find that tool at the following URL http://gvmtool.net/. You start by installing GVM, which has only two dependencies, namely ‘curl’ and ‘unzip’. These are essential-- if you don't have them installed then you cannot proceed with GVM. To be sure you can certainly install grails without GVM if you like. Check the instructions here.

Provided that you want to proceed with GVM, use the following command line evocation:

 curl -s get.gvmtool.net | bash

Once this line runs successfully the rest of the install becomes pretty easy.

Install Grails

with GVM in place installing grails should consist of exactly one line:

gvm install grails 2.4.3

Once this command completes successfully you should be able to run the following command from the command line (Note the extra - that the analogous Java command does not require):

grails –-version

and you should see a response that indicates that grails is working and running the version number you requested

A quick note for Mac users: homebrew also provides an easy approach to perform the Grails installation. And while we're on the topic installing Grails by hand is also quite easy. You download the package and set a GRAILS_HOME environmental variable and you're all set.

Install Groovy

Not strictly necessary (since grails comes with its own version of groovy built-in) but you may as well having come this far

gvm install groovy

If your Groovy installation completed successfully then the following command will tell you that Groovy is running

groovy --version

To be clear, you can run grails without installing Groovy explicitly. If you do install Groovy explicitly, however, then you can begin writing programs in straight up Groovy as well, which has its own attractions. Groovy is a really fun language with lots of possibilities. Check out the documentation here

Get the code

Now it’s time to pull down the code from the repository. Provided that you have already installed a git client this can be accomplished with one line

git clone [email protected]:broadinstitute/dig-diabetes-portal.git

Git will create a new directory called dig-diabetes-portal. Change your current working directory and then you will be ready to start building the system.

While your cloning repos note that both Groovy and Grails have their own repo, and that you can download the source code for either or both languages. Each of these repos is instructive and worth going through, especially if you get stuck somewhere along the way. For the record, note that Groovy is written mostly in Java (though there is currently a project underway to rewrite Groovy in Groovy). Grails, by comparison, is written mostly in Groovy.

Testing

A good place to start might be with running the tests. You can run the unit tests with the following command:

grails test-app unit:

You can run integration tests with this command:

grails test-app integration:

Skipping the 'unit' or 'integration' specification will cause the entire test suite to run. Conversely adding a class name to the end of one of the earlier commands will allow you to run only a single test.

Run the portal

Once you have demonstrated that you can run both tests to completion you can be reasonably confident that your environment is in good shape and that you are ready to run the portal. The command looks like this:

grails run-app

this command should be all you need to start a running application from the command line. If you'd like to do some development, however, you will probably want to import the project into an IDE. I like IntelliJ, though there is also an active community of Grails developers who use eclipse. If you would like to prepare your project for an IDE then there is a grails command for that. To open up the project in IntelliJ, for example, the command is:

grails integrate-with --intellij

which will generate an "*.ipr" file along with everything else you need to open the fully functional project under IntelliJ

If you want to now deploy a the portal you would need to use grails to create a WAR file that you could then the handoff to a suitable servlet container such as Apache or Tomcat. The grails command necessary to create a war file by hand is:

grails war

In order to increase consistency a bash script exists to build war files. This script is named 'gpw', and it passes a few parameters in as the war is created making the resulting executable easier to track. To run this script enter the following command from a bash-aware command prompt:

./gpw nameForThisVersion

Note that this script performs a 'git rev-parse HEAD' in order to determine your current git repo version, so your command line interface needs to recognize 'git'.

Configuration

While the diabetes portal is open source at its core, there are a few keys ( mostly relevant to authentication and services we pay for) which need to remain secret. As well, you may choose to override some of the default values utilized by the portal during compilation. In both cases you'll need to utilize a private configuration file that is stored somewhere on a local disk. The portal is looking for such a private configuration file. To determine the directory that should hold your private config, try running the above 'gpw' command and watch the resulting console output to find a line that look something like this:

>>>>>>>>>>>Note to developers: config files may be placed in the directory  = /Users/ben/.grails/dig-diabetes-portal

This line will tell you where on disk to store your private configuration file. In that directory you may then create a file named 'dig-diabetes-portal-commons-config.groovy'. This file will now be read in by grails during the compilation phase, and any values listed in this file will override those listed in the portals default configuration file (named Config.groovy). Any values not explicitly overridden will retain their default values. You will know that you have created a personal configuration file in the right place and with the right name if you see lines similar to the following when you compile the portal using gpw.

\*\* !! config override is in effect !! \*\*
\!\!\!\!\! file:/Users/ben/.grails/dig-diabetes-portal/dig-diabetes-portal-commons-config.groovy !! **

Note that a personal configuration file will (very) soon become mandatory, and that this file will need to contain at a minimum the following line:

oauth.providers.google.api.secret = 'xxxxxxx'

where xxxxxxx represents our Google OAuth secret key. For developers working on the diabetes portal in the Medical Population Genomics department at the Broad Institute then please see me (ba) for our current key. Otherwise set up your own authorization secret key with Google and replace xxxxxxx with your number.

Special note: if you don't create the above configuration file, or else if you put it in the wrong place then two things will happen: first, during compilation the Grails code will give you a polite, comprehensible message indicating that you did not supply a configuration file. That message will Show up on your console, and should look something like this:

\*\* No config override  in effect \*\*

as well, Grails will error out without running, and will provide a long and barely comprehensible error message that starts out looking like this:

Error |
Fatal error running tests: Error creating bean with name 'grails.plugin.databasemigration.DbdocController': 
Initialization of bean failed; nested exception is org.springframework.beans.factory.BeanCreationException: 
Error creating bean with name 'instanceControllerTagLibraryApi': 
Injection of autowired dependencies failed; nested exception is org.springframework.beans.factory.BeanCreationException: Could not autowire method: 
public void org.codehaus.groovy.grails.plugins.web.api.ControllerTagLibraryApi.setTagLibraryLookup(org.codehaus.groovy.grails.web.pages.TagLibraryLookup); 
nested exception is org.springframework.beans.factory.BeanCreationException: 
Error creating bean with name 'gspTagLibraryLookup': 
Invocation of init method failed; 
nested exception is org.springframework.beans.factory.BeanCreationException: 
Error creating bean with name 'grails.plugin.springsecurity.oauth.SpringSecurityOAuthTagLib': 
Initialization of bean failed; nested exception is org.springframework.beans.factory.BeanCreationException: 
Error creating bean with name 'oauthService': 
Invocation of init method failed; nested exception is java.lang.IllegalStateException: 
Missing oauth secret or key (or both!) in configuration for google. (Use --stacktrace to see the full trace)
.Tests FAILED 
|

So if you see the above message, look to see if you have a configuration file, if it is in the right place, and if it contains the right values.

Development tools:

Code coverage

```bash grails test-app -coverage ``` and see ./target/test-reports/cobertura

Automated code Checker

```bash grails codenarc ``` and find your summary report in ./target/CodeNarcReport.html

CI system and version promotion

We have a process to move from code push to production deploy. It looks like this:

  • 1)every time a developer pushes code:
  • 1.1) the CI system pulls the latest code from Git on the master branch
  • 1.2) the system runs all the unit and integration tests. IF they pass then
  • 1.3) CI deploys, and gives that version a CI Git tag. Then…
  • 2) every morning at 2 AM
  • 2.1) the tag describing last successfully deployed CI version is pulled from Git
  • 2.2) the system runs all the unit and integration tests. IF they pass then
  • 2.3) DEV deploys, and gives that version a DEV Git tag.
  • 2.4) developers(and potentially testers) look at the deployed DEV system, and decide whether they like it. If they do then eventually…
  • 3) a developer goes to the Jenkins project named 'MANUAL-deployedToQA'
  • 3.1) the developer goes in with the name of a DEV tag in hand. The developer chooses that tag from the drop-down list, then presses the 'Build' button
  • 3.2) the system runs all the unit and integration tests. IF they pass then
  • 3.3) the system BRANCHES the code, and deploys that branch to QA
  • 3.4) developers look at the system themselves, but also request the attention of testers/internal-users to see if this system is acceptable. If it is then...
  • 4) a developer goes to the Jenkins project named 'MANUAL-deployedToProd' (hopefully during nonpeak hours)
  • 4.1) the developer goes in with the name of a QA branch in hand. The developer chooses that branch name from the drop-down list, then presses the 'Build' button
  • 4.2) the system runs all the unit and integration tests. IF they pass then
  • 4.3) the system BRANCHES the code again, and deploys that branch to PROD
  • 5) and, as necessary...
  • 5.1) someone finds a critical bug that can't wait for the next iteration. A developer checks out the branch currently running on prod and fixes the problem.
  • 5.2) the developer checks in the fix. Since it isn't on the master branch it doesn't enter the CI system, so test carefully!
  • 5.3) tag your version by hand.
  • 5.4) Then follow steps (4) above to deploy the revised code
  • 5.5) Don't forget to merge your version back into the master branch before the next iteration begins!
done. Code has been deployed to PROD.

Note: This entire process should be completed once per iteration (or more, if any problems are discovered after the deployment). This deployment should happen at least a couple of days before the user group, so that the group can help take a look at the newly minted production code.

Install Jenkins Server

eg make install JENKINS_HOME=/humgen/diabetes/portal/dig-jenkins/test_server/home/ creates log in $JENKINS_HOME/Jenkins_${JENKINS_PORT}.log