Setup Guide¶
In this guide you learn how to setup the development environment of Artemis. Artemis is based on JHipster, i.e. Spring Boot development on the application server using Java 16, and TypeScript development on the application client in the browser using Angular and Webpack. To get an overview of the used technology, have a look at the JHipster Technology stack and other tutorials on the JHipster homepage.
You can find tutorials how to setup JHipster in an IDE (IntelliJ IDEA Ultimate is recommended) on https://jhipster.github.io/configuring-ide. Note that the Community Edition of IntelliJ IDEA does not provide Spring Boot support (see the comparison matrix). Before you can build Artemis, you must install and configure the following dependencies/tools on your machine:
Java JDK: We use Java (JDK 16) to develop and run the Artemis application server which is based on Spring Boot.
MySQL Database Server 8: Artemis uses Hibernate to store entities in a MySQL database. Download and install the MySQL Community Server (8.0.x) and configure the ‘root’ user with an empty password. (In case you want to use a different password, make sure to change the value in application-dev.yml and in liquibase.gradle). The required Artemis scheme will be created / updated automatically at startup time of the server application. Alternatively, you can run the MySQL Database Server inside a Docker container using e.g.
docker-compose -f src/main/docker/mysql.yml up
Node.js: We use Node LTS (>=14.15.0 < 15) to compile and run the client Angular application. Depending on your system, you can install Node either from source or as a pre-packaged bundle.
Yarn: We use Yarn (>=1.22.4 <2) to manage client side Node dependencies. Depending on your system, you can install Yarn either from source or as a pre-packaged bundle. To do so, please follow the instructions on the Yarn installation page.
Server Setup¶
To start the Artemis application server from the development
environment, first import the project into IntelliJ and then make sure
to install the Spring Boot plugins to run the main class
de.tum.in.www1.artemis.ArtemisApp
. Before the application runs, you
have to configure the file application-artemis.yml
in the folder
src/main/resources/config
.
artemis:
repo-clone-path: ./repos/
repo-download-clone-path: ./repos-download/
encryption-password: <encrypt-password> # arbitrary password for encrypting database values
user-management:
use-external: true
external:
url: https://jira.ase.in.tum.de
user: <username> # e.g. ga12abc
password: <password>
admin-group-name: tumuser
ldap:
url: <url>
user-dn: <user-dn>
password: <password>
base: <base>
version-control:
url: https://bitbucket.ase.in.tum.de
user: <username> # e.g. ga12abc
password: <password>
token: <token> # VCS API token giving Artemis full Admin access. Not needed for Bamboo+Bitbucket
ci-token: <token from the CI> # Token generated by the CI (e.g. Jenkins) for webhooks from the VCS to the CI. Not needed for Bamboo+Bitbucket
continuous-integration:
url: https://bamboo.ase.in.tum.de
user: <username> # e.g. ga12abc
token: <token> # Enter a valid token generated by bamboo or leave this empty to use the fallback authentication user + password
password: <password>
vcs-application-link-name: LS1 Bitbucket Server # If the VCS and CI are directly linked (normally only for Bitbucket + Bamboo)
empty-commit-necessary: true # Do we need an empty commit for new exercises/repositories in order for the CI to register the repo
# Hash/key of the ci-token, equivalent e.g. to the ci-token in version-control
# Some CI systems, like Jenkins, offer a specific token that gets checked against any incoming notifications
# from a VCS trying to trigger a build plan. Only if the notification request contains the correct token, the plan
# is triggered. This can be seen as an alternative to sending an authenticated request to a REST API and then
# triggering the plan.
# In the case of Artemis, this is only really needed for the Jenkins + GitLab setup, since the GitLab plugin in
# Jenkins only allows triggering the Jenkins jobs using such a token. Furthermore, in this case, the value of the
# hudson.util.Secret is stored in the build plan, so you also have to specify this encrypted string here and NOT the actual token value itself!
# You can get this by GETting any job.xml for a job with an activated GitLab step and your token value of choice.
secret-push-token: <token hash>
# Key of the saved credentials for the VCS service
# Bamboo: not needed
# Jenkins: You have to specify the key from the credentials page in Jenkins under which the user and
# password for the VCS are stored
vcs-credentials: <credentials key>
# Key of the credentials for the Artemis notification token
# Bamboo: not needed
# Jenkins: You have to specify the key from the credentials page in Jenkins under which the notification token is stored
notification-token: <credentials key>
# The actual value of the notification token to check against in Artemis. This is the token that gets send with
# every request the CI system makes to Artemis containing a new result after a build.
# Bamboo: The token value you use for the Server Notification Plugin
# Jenkins: The token value you use for the Server Notification Plugin and is stored under the notification-token credential above
authentication-token: <token>
lti:
id: artemis_lti
oauth-key: artemis_lti_key
oauth-secret: <secret> # only important for online courses on the edX platform, can typically be ignored
user-prefix-edx: edx_
user-prefix-u4i: u4i_
user-group-name-edx: edx
user-group-name-u4i: u4i
git:
name: Artemis
email: artemis@in.tum.de
athene:
url: http://localhost
base64-secret: YWVuaXF1YWRpNWNlaXJpNmFlbTZkb283dXphaVF1b29oM3J1MWNoYWlyNHRoZWUzb2huZ2FpM211bGVlM0VpcAo=
token-validity-in-seconds: 10800
Change all entries with <...>
with proper values, e.g. your TUM
Online account credentials to connect to the given instances of JIRA,
Bitbucket and Bamboo. Alternatively, you can connect to your local JIRA,
Bitbucket and Bamboo instances. It’s not necessary to fill all the
fields, most of them can be left blank. Note that there is additional
information about the setup for programming exercises provided:
Note
Be careful that you don’t commit changes to application-artemis.yml
.
To avoid this, follow the best practice when configuring your local development environment:
Create a file named
application-local.yml
undersrc/main/resources/config
.Copy the contents of
application-artemis.yml
into the new file.Update configuration values in
application-local.yml
.
By default, changes to application-local.yml
will be ignored by git so you don’t accidentally
share your credentials or other local configuration options.
If you use a password, you need to adapt it in
gradle/liquibase.gradle
.
Run the server via a service configuration¶
This setup is recommended for production instances as it registers Artemis as a service and e.g. enables auto-restarting of Artemis after the VM running Artemis has been restarted. For development setups, see the other guides below.
This is a service file that works on Debian/Ubuntu (using systemd):
[Unit]
Description=Artemis
After=syslog.target
[Service]
User=artemis
WorkingDirectory=/opt/artemis
ExecStart=/usr/bin/java \
-Djdk.tls.ephemeralDHKeySize=2048 \
-DLC_CTYPE=UTF-8 \
-Dfile.encoding=UTF-8 \
-Dsun.jnu.encoding=UTF-8 \
-Djava.security.egd=file:/dev/./urandom \
-Xmx2048m \
--add-modules java.se \
--add-exports java.base/jdk.internal.ref=ALL-UNNAMED \
--add-exports java.naming/com.sun.jndi.ldap=ALL-UNNAMED \
--add-opens java.base/java.lang=ALL-UNNAMED \
--add-opens java.base/java.nio=ALL-UNNAMED \
--add-opens java.base/sun.nio.ch=ALL-UNNAMED \
--add-opens java.management/sun.management=ALL-UNNAMED \
--add-opens jdk.management/com.sun.management.internal=ALL-UNNAMED \
-jar artemis.war \
--spring.profiles.active=prod,bamboo,bitbucket,jira,ldap,scheduling,openapi
SuccessExitStatus=143
StandardOutput=/opt/artemis/artemis.log
StandardError=inherit
[Install]
WantedBy=multi-user.target
The following parts might also be useful for other (production) setups, even if this service file is not used:
-Djava.security.egd=file:/dev/./urandom
: This is required if repositories are cloned via SSH from the VCS.The default (pseudo-)random-generator
/dev/random
is blocking which results in very bad performance when using SSH due to lack of entropy.
The file should be placed at /etc/systemd/system/artemis.service
and after running sudo systemctl daemon-reload
, you can start the service using sudo service artemis start
.
You can stop the service using sudo service artemis stop
and restart it using sudo service artemis restart
.
Logs can be fetched using sudo journalctl -u artemis -f -n 200
.
Run the server via a run configuration in IntelliJ¶
The project comes with some pre-configured run / debug configurations that are stored in the .idea
directory.
When you import the project into IntelliJ the run configurations will also be imported.
The recommended way is to run the server and the client separated. This provides fast rebuilds of the server and hot module replacement in the client.
Artemis (Server): The server will be started separated from the client. The startup time decreases significantly.
Artemis (Client): Will execute
yarn install
andyarn start
. The client will be available at http://localhost:9000/ with hot module replacement enabled (also see Client Setup).
Other run / debug configurations¶
Artemis (Server & Client): Will start the server and the client. The client will be available at http://localhost:8080/ with hot module replacement disabled.
Artemis (Server, Jenkins & Gitlab): The server will be started separated from the client with the profiles
dev,jenkins,gitlab,artemis
instead ofdev,bamboo,bitbucket,jira,artemis
.Artemis (Server, Athene): The server will be started separated from the client with
athene
profile enabled (see Athene Service).
Typical problems with Liquibase checksums¶
One typical problem in the development setup is that an exception occurs during the database initialization. Artemis uses Liquibase to automatically upgrade the database scheme after changes to the data model. This ensures that the changes can also be applied to the production server. In case you encounter errors with liquibase checksum values, run the following command in your terminal / command line:
./gradlew liquibaseClearChecksums
Run the server with Spring Boot and Spring profiles¶
The Artemis server should startup by running the main class
de.tum.in.www1.artemis.ArtemisApp
using Spring Boot.
Note
Artemis uses Spring profiles to segregate parts of the
application configuration and make it only available in certain
environments. For development purposes, the following program arguments
can be used to enable the dev
profile and the profiles for JIRA,
Bitbucket and Bamboo:
--spring.profiles.active=dev,bamboo,bitbucket,jira,artemis,scheduling
If you use IntelliJ (Community or Ultimate) you can set the active profiles by
Choosing
Run | Edit Configurations...
Going to the
Configuration Tab
Expanding the
Environment
section to revealVM Options
and setting them to-Dspring.profiles.active=dev,bamboo,bitbucket,jira,artemis,scheduling
Set Spring profiles with IntelliJ Ultimate¶
If you use IntelliJ Ultimate, add the following entry to the section
Active Profiles
(within Spring Boot
) in the server run
configuration:
dev,bamboo,bitbucket,jira,artemis,scheduling
Run the server with the command line (Gradle wrapper)¶
If you want to run the application via the command line instead, make
sure to pass the active profiles to the gradlew
command like this:
./gradlew bootRun --args='--spring.profiles.active=dev,bamboo,bitbucket,jira,artemis,scheduling'
As an alternative, you might want to use Jenkins and Gitlab with an internal user management in Artemis, then you would use the profiles:
dev,jenkins,gitlab,artemis,scheduling
Client Setup¶
You need to install Node and Yarn on your local machine.
Using IntelliJ¶
If you are using IntelliJ you can use the pre-configured Artemis (Client)
run configuration that will be delivered with this repository:
Choose
Run | Edit Configurations...
Select the
Artemis (Client)
configuration from thenpm section
Now you can run the configuration in the upper right corner of IntelliJ
Using the command line¶
You should be able to run the following
command to install development tools and dependencies. You will only
need to run this command when dependencies change in package.json
.
yarn install
To start the client application in the browser, use the following command:
yarn start
This compiles TypeScript code to JavaScript code, starts the hot module
replacement feature in Webpack (i.e. whenever you change a TypeScript
file and save, the client is automatically reloaded with the new code)
and will start the client application in your browser on
http://localhost:9000
. If you have activated the JIRA profile (see
above in Server Setup) and if you have configured
application-artemis.yml
correctly, then you should be able to login
with your TUM Online account.
In case you encounter any problems regarding JavaScript heap memory leaks when executing yarn start
or any other scripts from package.json
, you can change the memory limit parameter used in the webpack-ts
script in package.json
to the following:
# This local change in `package.json` should not be committed.
"webpack-ts": "cross-env NODE_OPTIONS=--max_old_space_size=5120 TS_NODE_PROJECT=\"tsconfig.webpack.json\" webpack" # possible higher values are 6144, 7168, and 8192
Make sure to not commit this change in package.json
.
For more information, review Working with Angular. For further instructions on how to develop with JHipster, have a look at Using JHipster in development.
Customize your Artemis instance¶
You can define the following custom assets for Artemis to be used instead of the TUM defaults:
The logo next to the “Artemis” heading on the navbar →
${artemisRunDirectory}/public/images/logo.png
The favicon →
${artemisRunDirectory}/favicon.svg
The privacy statement HTML →
${artemisRunDirectory}/public/content/privacy_statement.html
The imprint statement HTML →
${artemisRunDirectory}/public/content/imprint.html
The contact email address in the
application-{dev,prod}.yml
configuration file under the keyinfo.contact
Alternative: Using docker-compose¶
A full functioning development environment can also be set up using docker-compose:
Install docker and docker-compose
Configure the credentials in
application-artemis.yml
in the foldersrc/main/resources/config
as described aboveRun
docker-compose up
Go to http://localhost:9000
The client and the server will run in different containers. As yarn is
used with its live reload mode to build and run the client, any change
in the client’s codebase will trigger a rebuild automatically. In case
of changes in the codebase of the server one has to restart the
artemis-server
container via
docker-compose restart artemis-server
.
(Native) Running and Debugging from IDEs is currently not supported.
Get a shell into the containers:¶
app container:
docker exec -it $(docker-compose ps -q artemis-app) sh
mysql container:
docker exec -it $(docker-compose ps -q artemis-mysql) mysql
Other useful commands:¶
Stop the server:
docker-compose stop artemis-server
(restart viadocker-compose start artemis-server
)Stop the client:
docker-compose stop artemis-client
(restart viadocker-compose start artemis-client
)
Athene Service¶
The semi-automatic text assessment relies on the Athene service. To enable automatic text assessments, special configuration is required:
Enable the athene
Spring profile:¶
--spring.profiles.active=dev,bamboo,bitbucket,jira,artemis,scheduling,athene
Configure API Endpoints:¶
The Athene service is running on a dedicated machine and is addressed via
HTTP. We need to extend the configuration in the file
src/main/resources/config/application-artemis.yml
like so:
artemis:
# ...
athene:
url: http://localhost
base64-secret: YWVuaXF1YWRpNWNlaXJpNmFlbTZkb283dXphaVF1b29oM3J1MWNoYWlyNHRoZWUzb2huZ2FpM211bGVlM0VpcAo=
token-validity-in-seconds: 10800
Athene Service¶
The semi-automatic text assessment relies on the Athene service. To enable automatic text assessments, special configuration is required:
Enable the apollon
Spring profile:¶
--spring.profiles.active=dev,bamboo,bitbucket,jira,artemis,scheduling,apollon
Configure API Endpoints:¶
The Apollon conversion service is running on a dedicated machine and is adressed via
HTTP. We need to extend the configuration in the file
src/main/resources/config/application-artemis.yml
like so:
apollon:
conversion-service-url: http://localhost:8080