This is not an official Google product.

Node-RED nodes for Google Cloud Platform.

Node-RED is an open source project which provides a visual programming environment and runtime for execution of event based applications. Written in Node.js JavaScript, it executes on a wide variety of platforms including workstations, IoT devices and, most importantly for our discussion, on Google Cloud Platform environments which includes Compute Engines and Kubernetes.

The Node-RED story is based on discrete nodes of function which are dragged and dropped from a palette to a canvas and then wired together. This resulting flow of execution is then triggered by the arrival of external events including REST requests, Pub/Sub messages, timers and more.

One of the key reasons that Node-RED has become as popular as it has is the ease with which developers can build additional nodes that encapsulate rich sets functions. Once written, these add-on nodes can be used by flow writers without having to know the complexities of their underlying operation. One just drags a new node onto the canvas and uses it.

This project provides a set of core Google Cloud Platform oriented Node-RED nodes. Once installed, the capability to use a broad set of GCP services are immediately unlocked.

There are multiple ways to install node-red-contrib-google-cloud. The official ways are described in the Node-RED documentation. The name of the package for installation is node-red-contrib-google-cloud.

Each of the new nodes made available through this package will communicate with the Google Cloud Platform (GCP). These interactions must be performed securely and require authentication information to be passed.

If Node-RED is running under a GCP environment such as a Compute Engine, Google Kubernetes Engine or Cloud Run then there is an implicit identity presented to GCP and the flow developer need do nothing special for authentication configuration. However, if your Node-RED runtime is not running under GCP or you wish to call a service with adistinct identity, then you will need to use explicit credentials.

On each node, we have the opportunity to supply credentials. These can be supplied either as a path to a named key file or by creating a Node-RED managed name credentials secret. Each credential defined as a Node-RED secret has the following properties:

The credentials for a service account can be acquired from the APIs & Services menu. After you finish creating a service account key, it will be downloaded in JSON format and saved to a local file. Copy and paste the contents of the file directly into the Key field in the node editor.

An alternative to supplying credentials through the Node-RED credentials mechanism is to point individual GCP nodes to a JSON key file that exists on the file system on which Node-RED is running. This capability was introduced to support the Kubernetes "secrets" architecture. A Kubernetes administrator could obtain a JSON key file and save that as a secret in the Kubernetes environment. Containers within the environment could then see the secrets as mounted files. This level of indirection and abstraction allows us to build images which don't contain the keys hard-coded into the images but yet make those keys available at run-time. In addition, if we need to change or rotate the keys, we can do so without having to rebuild any images.

When we run Node-RED on GCP compute resources such as Compute Engine or GKE, the environment to make GCP API service calls is already present. If we run Node-RED outside of GCP (for example on a desktop PC, an on-premises server or a Raspberry Pi) then some additional setup to connect and use GCP APIs is required. Specifically, an environment variable called GOOGLE_CLOUD_PROJECT must be set and be present in the environment in which Node-RED runs. The value of this variable should be the GCP project that you are going to interact with.

If you are running Node-RED from the command line, you can use:

export GOOGLE_CLOUD_PROJECT=<YourProjectId>
node-red

You can also permanently set the variable by editing /etc/environment and adding a line which reads:

GOOGLE_CLOUD_PROJECT=<YourProjectId>

Note: For Raspberry Pi users - There is the option of starting Node-RED automatically through the Linux systemctl daemon. This system does not use global environment variables and thus you must explicitly define the variable to use.

Edit the file /lib/systemd/system/nodered.service

and add a line which reads:

Environment="GOOGLE_CLOUD_PROJECT=<YourProjectId>"

within the [Service] section of the file.

The set of Node-RED nodes are found in the GCP section of the palette. The current set of nodes are:

• PubSub In - Receive a published message.
• PubSub Out - Publish a new message.
• Google Cloud Store Read - Read from the object store.
• Google Cloud Store Write - Write to the object store.
• Log - Write an entry to Stackdriver logging.
• Tasks - Create a new Cloud Task instance.
• Entities - Perform Natural Language analysis of entities on a piece of text.
• Sentiment - Perform Natural Language analysis of sentiment on a piece of text.
• Vision - Analyse an image.
• Metadata - Retrieve the compute engine metadata.
• Monitoring - Write a new monitoring record.
• Speech to Text - Recognize speech from audio.
• Text to Speech - Encode text as spoken audio.
• Data loss prevention - Mask out identifiable information.
• Translation - Translate text from one language to another.
• BigQuery - Execute a BigQuery query.
• BigQuery Insert - Execute a BigQuery insert.
• IoT Command Send - Send an IoT command to a device.
• IoT Config Update - Update the config of an IoT device.
• IoT Telemetry Send - Send a telemetry message from a device.
• AutoML - Execute an AutoML prediction.
• DNS Change Create
• DNS Records List
• GCS Signed URL
• Document AI - Parse the contents of a PDF or image of a document.
• Spanner - Execute a SQL statement against Spanner.

We are very open to receiving feedback on additional GCP nodes that may be of value. Don't hesitate to open an issue should you have a desire for incorporating additional GCP functions.

• Blog: Low code programming with Node-RED comes to GCP - 2020-02
• Medium: GCP Document AI and Node-RED - 2021-02
• Medium: Node-RED solution deployment on GCP - 2021-03