| commit | 3f36fb207755a239ffde933988f6d6e233716b71 | [log] [tgz] |
|---|---|---|
| author | avigaffa <avi.gaffa@amdocs.com> | Mon Dec 31 10:15:56 2018 +0200 |
| committer | avigaffa <avi.gaffa@amdocs.com> | Mon Dec 31 10:20:48 2018 +0200 |
| tree | 70d4969321cbde7c256767d2d2b420b1a90d561a | |
| parent | 3a2ab7d1afa973b1d10dc12d78eb65b8d1193bcd [diff] |
upgrade io.fabric8 docker-maven-plugin version Spliting install in docker file to 2 rows to see where the build is failing on ARM Issue-ID: SDC-2026 Change-Id: I51fd800bd368d2eb1050d5b412fc15b8cd5b1ea6 Signed-off-by: avigaffa <avi.gaffa@amdocs.com>
Workflow Designer is a pluggable SDC designer that allows a user to design a workflow, save it, and attach it to a SDC service as an artifact. Workflow Designer also manages the definitions of activities, which can be later used as parts of the designed workflows.
The designer is comprised of the following deployment units:
Designer backend is the core component. It exposes RESTful APIs for managing workflow and activity data. The backend is agnostic to the type of a workflow artifact — its main concerns are workflow inputs and outputs, and metadata. One of the APIs enables to attach a certified workflow artifact to a SDC service, therefore the designer must be able to call an API on SDC. In order to do so, the location of a SDC server, and SDC consumer credentials are required.
Designer frontend serves static content of a Web application for creating and managing workflows, and forwards API requests to the backend. The static content includes JavaScript, images, CSS, etc. A major part of the Web application is Workflow Composition View — a graphical interface for arranging a workflow sequence. The Web application also produces a workflow artifact that will be sent to the backend, saved along with other data, and later used by a service. The architecture allows for different implementations of the frontend component. For example, a different technology can be used for the Composition View, which will probably also result in a different type of the artifacts (e.g. Bpmn.io vs. Camunda).
Cassandra database is used by the designer backend as the main storage for workflow data. A dedicated instance of Cassandra can be deployed, or an existing cluster may be used.
Database initialization scripts run once per deployment to create the necessary Cassandra keyspaces and tables, pre-populate data, etc.
The procedure below describes manual deployment on plain Docker for development or a demo.
Create a dedicated instance of Cassandra. This step is optional if you already have a Cassandra cluster. The designer is not expected to have problems working with Cassandra 3.x, but has been tested with 2.1.x because this is the version used by SDC.
An easy way to spin up a Cassandra instance is using a Cassandra Docker image as described in the official documentation.
docker run -d --name workflow-cassandra cassandra:2.1
WARNING: This step must be executed only once.
docker run -ti -e CS_HOST=<cassandra-host> -e CS_PORT=<cassandra-port> -e CS_AUTHENTICATE=true/false -e CS_USER=<cassandra-user> -e CS_PASSWORD=<cassandra-password> nexus3.onap.org:10001/onap/workflow-init:latest
CS_HOST — Cassandra hostname or IP address.
CS_PORT — Cassandra Thrift client port. If not specified, the default of 9160 will be used.
CS_AUTHENTICATE — whether password authentication must be used to connect to Cassandra. A false will be assumed if this variable is not specified.
CS_USER — Cassandra username if CS_AUTHENTICATE is true.
CS_PASSWORD — Cassandra password if CS_AUTHENTICATE is true.
Assuming you have created a dedicated Cassandra container as described in Database section, and the access to it is not protected with a password, the following command will initialize the database:
docker run -d --name workflow-init -e CS_HOST=$(docker inspect workflow-cassandra --format={{.NetworkSettings.IPAddress}}) nexus3.onap.org:10001/onap/workflow-init:latest
In order to see if the Workflow Designer was successfully initialized, make sure the console does not contain error messages. You can also see the logs of the initialization container using docker logs workflow-init command.
docker run -d -e SDC_PROTOCL=http/https -e SDC_ENDPOINT=<sdc-host>:<sdc-port> -e SDC_USER=<sdc-username> -e SDC_PASSWORD=<sdc-password> -e CS_HOSTS=<cassandra-hosts> -e CS_PORT=<cassandra-port> -e CS_AUTHENTICATE=true/false -e CS_USER=<cassandra-user> -e CS_PASSWORD=<cassandra-password> -e JAVA_OPTIONS=<jvm-options> nexus3.onap.org:10001/onap/workflow-backend:latest
SDC_PROTOCOL — protocol to be used for calling SDC APIs (http or https).
SDC_ENDPOINT — the base path of SDC external API, in the format host:port, where host is a SDC backend server, and port is usually 8080.
SDC_USER — Workflow consumer username
SDC_PASSWORD — Workflow consumer password
CS_HOSTS — comma-separated list of Cassandra hostnames or IP addresses.
CS_PORT — CQL native client port. If not specified, the default of 9042 will be used.
CS_AUTHENTICATE — whether password authentication must be used to connect to Cassandra. A false will be assumed if this variable is not specified.
CS_USER — Cassandra username if CS_AUTHENTICATE is true.
CS_PASSWORD — Cassandra password if CS_AUTHENTICATE is true.
JAVA_OPTIONS — optionally, JVM (Java Virtual Machine) arguments.
Assuming you have a dedicated Cassandra container as described in Database section, and the access to it is not protected with a password. The following command will start a backend container:
docker run -d --name workflow-backend -e SDC_PROTOCOL=http -e SDC_ENDPOINT=$(docker inspect sdc-BE --format={{.NetworkSettings.IPAddress}}):8080 -e CS_HOSTS=$(docker inspect workflow-cassandra --format={{.NetworkSettings.IPAddress}}) -e SDC_USER=workflow -e SDC_PASSWORD=<secret> -e JAVA_OPTIONS="-Xmx128m -Xms128m -Xss1m" nexus3.onap.org:10001/onap/workflow-backend:latest
In order to verify that the Workflow Designer backend has started successfully, check the logs of the backend container. For example, by running docker logs workflow-backend. The logs must not contain any error messages.
Application logs are located in the /var/log/ONAP/workflow-designer/backend directory of a workflow backend container. For example, you can view the audit log by running docker exec -ti workflow-backend less /var/log/ONAP/workflow-designer/backend/audit.log.
docker run -d -e BACKEND=http://<backend-host>:<backend-port> -e JAVA_OPTIONS=<jvm-options> nexus3.onap.org:10001/onap/workflow-frontend:latest
BACKEND — root endpoint of the RESTful APIs exposed by a workflow backend server.
JAVA_OPTIONS — optionally, JVM (Java Virtual Machine) arguments.
docker run -d --name workflow-frontend -e BACKEND=http://$(docker inspect workflow-backend --format={{.NetworkSettings.IPAddress}}):8080 -e JAVA_OPTIONS="-Xmx64m -Xms64m -Xss1m" -p 9088:8080 nexus3.onap.org:10001/onap/workflow-frontend:latest
Notice that port 8080 of the frontend container has been mapped to port 9088 of the host machine. This makes the Workflow Designer Web application accessible from the outside world via the host machine's IP address/hostname.
In order to check if the Workflow Designer frontend has successfully started, look at the logs of the frontend container. For example, by running docker logs workflow-frontend. The logs should not contain error messages.
Workflow frontend does not have backend logic, therefore there are no application logs.
Docker Compose further simplifies the deployment of Workflow Designer. The Docker Compose files can be find in the workflow designer Git repository, under docker-compose directory.
In order to use this deployment method, you need to install Docker Compose as described on Install Docker Compose official page.
Instantiation of a Cassandra database is not part of the the Docker Compose service. You may already have a running instance of Cassandra you want to use. It can be in a Docker container, on a VM, or a physical machine.
If you want to spin up a Cassandra database alongside the Workflow Designer service for development purposes, use the following command:
docker-compose -p workflow -f cassandra.yml up -d
Note, that since the database was created under the same project (
-p workflow), but as a separate service, it will keep running when you shut down the workflow designer service. This will cause an error message ERROR: network workflow_default id <......> has active endpoints.
Edit .env file located in docker-compose directory. Here is a brief overview of some variables.
IMAGE_TAG — enables to try other versions of the Docker images
REGISTRY — allows to use any Docker registry; leave it blank for locally built images
CS_HOST — Cassandra host name or IP address. Keep in mind that the host must be accessible from the Docker Compose network created for the workflow service. Use CS_HOST=cassandra if you created the database as described in the previous section.
SDC_HOST — usually, IP address of the Docker host (if you are using the SDC deploy script).
CASSANDRA_INIT_PORT — Cassandra Thrift port, usually 9160.
CASSANDRA_PORT — Cassandra CQL native client port, usually 9042.
BACKEND_DEBUG_PORT — host port used to debug the backend server (see below).
FRONTEND_DEBUG_PORT — host port used to debug the frontend server (see below).
FRONTEND_PORT — host port Workflow Designer UI will be accessible at.
Other variables are described in Deployment on Docker section.
Assuming the database is up and running, execute the following in the command line:
docker-compose -p workflow up -d.
It is easy to restart or recreate the entire service or a selected component using Docker Compose commands, for example to pick up new versions of the Docker images. Keep in mind that the database may remain unchanged, so that the new service will continue to work with old data.
For example, you can restart just the frontend by issuing the command:
docker-compose -p workflow restart workflow-frontend
Keep in mind that changes to the docker-compose.yml configuration or environment variables will not be reflected when using restart. For that, you will need to recreate the container (e.g. to change the image version):
docker-compose -p workflow up -d --no-deps workflow-frontend
For more advanced features and commands, please refer to Docker Compose documentation.
You can shut down the entire stack of Workflow Designer components using docker-compose -p workflow down.
It is possible to start the service in debug mode, when both the front-end and the back-end are accessible from a remote debugger at mapped host ports. Run:
docker-compose -p workflow -f docker-compose.yml -f debug.yml up -d.
In order to run as an SDC pluggable designer, Workflow Designer must be added to SDC configuration as described in Generic plugin support.
If you are deploying SDC using a standard procedure (OOM or the SDC shell script), the easiest way to configure the Workflow plugin is to edit the default_attributes/Plugins/WORKFLOW section of AUTO.json.
The main endpoint to load Workflow Designer Web application is defined by "pluginSourceUrl": "http://<host>:<port>".
Keep in mind that the URL must be accessible from a user's browser. In most cases, <host> will be the hostname or IP address of the machine that runs Docker engine, and <port> will be a host port to which you have published port 8080 of the Workflow frontend container.
In order to check the availability of a plugin, SDC uses "pluginDiscoveryUrl". For Workflow the value is http://<host>:<port>/ping.
Let's assume that hostname of the machine that runs Docker containers with the Workflow application is workflow.example.com, and port 8080 of the Workflow frontend is mapped to 9088 on the host. In this case the corresponding section of AUTO.json will look like below:
"Plugins": { "WORKFLOW": { "workflow_discovery_url": "http://workflow.example.com:9088/ping", "workflow_source_url": "http://workflow.example.com:9088" } },
In a development or demo environment, Workflow Designer will run on the same host as SDC, so that its IP address will be the one of the Docker host.