How can I remotely deploy a flow file in Node-RED?

Question

Is there any way for deploying an existing Node-RED flow file myflow.json to a remote machine running Node-RED?

After much googling, I stumbled upon this discussion in the relevant Google group, but it is not very enlightening.

UPDATE (after the suggested answer below)

Here is the output of the suggested request from a remote machine:

[scripts]$ curl -v -X POST http://192.168.70.73:1880/flows -H "Content-Type: application/json"  --data "@remote_test.json" 
* About to connect() to 192.168.70.73 port 1880 (#0)
*   Trying 192.168.70.73... connected
* Connected to 192.168.70.73 (192.168.70.73) port 1880 (#0)
> POST /flows HTTP/1.1
> User-Agent: curl/7.19.7 (x86_64-redhat-linux-gnu) libcurl/7.19.7 NSS/3.21 Basic ECC zlib/1.2.3 libidn/1.18 libssh2/1.4.2
> Host: 192.168.70.73:1880
> Accept: */*
> Content-Type: application/json
> Content-Length: 1088
> Expect: 100-continue
> 
< HTTP/1.1 100 Continue
< HTTP/1.1 204 No Content
< X-Powered-By: Express
< Date: Fri, 10 Feb 2017 11:45:24 GMT
< Connection: keep-alive
< 
* Connection #0 to host 192.168.70.73 left intact
* Closing connection #0

with a 204 response (indicating success).

Here is the respective output in the local machine console:

10 Feb 13:45:24 - [info] Stopping flows
10 Feb 13:45:24 - [info] Stopped flows
10 Feb 13:45:24 - [info] Starting flows
10 Feb 13:45:24 - [info] Started flows

after which nothing happens.

The flow-to-be-deployed remote_test.json does indeed show up in the GUI of the local machine, but it is not running; furthermore, when I try to manually inject or enable the debug node, I get an Error: node not deployed in the local GUI (and the 'Deploy' button is inactive).

This is the ultra-simple flow:

with the format node simply parsing the temperature measurements received from get_temp every 5 seconds and adding a timestamp:

var temp = parseFloat(msg.payload.replace( /[^\d\.]*/g, ''));
var time = new Date().toISOString().replace(/\..+/, '');

msg.payload = {'measurements' : [ { 'time'        : time }, 
                                  { 'temperature' : temp } ]
};
console.log(msg.payload);
return msg;

And here is the console output when the flow is deployed locally:

10 Feb 14:08:25 - [info] Stopped flows
10 Feb 14:08:25 - [info] Starting flows
10 Feb 14:08:25 - [info] Started flows
{ measurements: [ { time: '2017-02-10T12:08:30' }, { temperature: 34.7 } ] }
{ measurements: [ { time: '2017-02-10T12:08:35' }, { temperature: 34.2 } ] }
{ measurements: [ { time: '2017-02-10T12:08:40' }, { temperature: 33.6 } ] }
{ measurements: [ { time: '2017-02-10T12:08:45' }, { temperature: 33.6 } ] }

Starting the service with node-red-start or node-red, as well as including or not the argument -H "Node-RED-Deployment-Type: full" in the request, do not make any difference.

System configuration:

• Raspberry Pi 3
• Node.js 7.4.0
• Node-RED 0.13.4

UPDATE 2

It seems that the issue was with the Node.js and/or Node-RED versions. Tried it in a different Pi, running Node.js 6.9.5 & Node-RED 0.16.2, and it works OK.

Answer 1

The Node-RED runtime exposes a REST api that provides methods for deploying a new flow configuration on the runtime.

The specific API in question is documented here: http://nodered.org/docs/api/admin/methods/post/flows/

It allows you to do an HTTP POST containing the flow configuration you want the runtime to deploy.

For example, if flow file is in myflow.json and Node-RED is running at http://localhost:1880, you can use curl to deploy it:

curl -X POST http://localhost:1880/flows -H "Content-Type: application/json" --data "@myflow.json"