Skip to content

Infrastructure for sending real time messages to the user (frontend).

Notifications You must be signed in to change notification settings

LCOGT/photonranch-userstatus

Repository files navigation

Photon Ranch User Status

This is a Serverless project that deploys a websocket api used to send messages from an observatory to the frontend at www.photonranch.org in real time. It is reminiscent of a logging pipeline, and messages are presented as timestamped messages with a priority status (eg. debug, info, warn, error, or critical).

The original task was streaming logs from the observatory so operators could more easily debug/troubleshoot from just the frontend. But we've pivoted to using this as a way to send more general messages to the user instead. The original name was 'logstream', and the backend resources are still named with that convention in mind (even though the repository name has diverged).

Description

Observatory owners can post logs to describe in detail the processes occuring at an observatory for users to follow along with as they occur. These are accessible for a given timeframe, defined for our purposes as 24 hours.

Viewing a message as a user

Architecture

This Serverless app creates api gateway instances, one for http and one for websockets. These both route requests to a layer of lambda functions, which interact with two dynamodb tables: one for storing messages, and one for keeping track of websocket connections.

The messages table adds a TTL column which expires the messages after a few days. We decided that it is not worth the maintenance burden to archive all our messages; they are treated as transient notifications only.

A basic architecture diagram is provided below. Note that there are additional endpoints and lambda functions not described, but this tells the basic idea.

Architecture diagram showing get and post requests

Dependencies

Dependencies will be listed here.

Local Development

Clone the repository to your local machine:

git clone https://github.com/LCOGT/photonranch-userstatus.git
cd photonranch-userstatus

Requirements

You will need the Serverless Framework installed locally for development. For manual deployment to AWS as well as for updating dependencies, you will need to install Node, npm, and AWS CLI, configuring with your own AWS credentials.

Deployment

This project currently has two stages, prod and dev. For manual deployment on your local machine, you'll need to install packages:

npm install
serverless plugin install --name serverless-python-requirements

To deploy, run:

serverless deploy --stage {stage}

Testing

Instructions to manually run tests will be detailed here.

Event Syntax

There are four parameters supplied in the body of a message:

{
    "site": "tst",  // (str) code for the site that will display the message
    "log_message": "Slewing telescope...", // (str) content that the 
    // user will read.
    "log_level": "warning", //(str) can be ["debug", "info", "warning", "error", "critical"] 
    // following the python logging convention. Default (if none provided) is info.
    "timestamp": time.time(), //(int) unix timestamp in seconds. Messages are sorted and 
    // displayed chronologically using this value; the hh:mm time prefixes the message display.
}

Websocket messages will arrive with the following structure:

{
    "site":"tst",
    "message":"This is a log message for testing.\nIt is sent from the frontend.",
    "log_level":"warning",
    "timestamp":"1604514063"
}

API Endpoints

Depending on your setup, the Serverless project will deploy with a unique endpoint url. For Photon Ranch, we will use the base url https://logs.photonranch.org/logs. An additional dev endpoint is located at https://logs.photonranch.org/dev.

  • POST /newlog

    • Description: Add new message entry in the log.
    • Authorization required: No
    • Request body: JSON body as specified in syntax above.
    • Responses:
      • 200: Successfully added message
    • Example request:
    # python 3.6
    import time, reqeusts, json
    def send_log_to_frontend():
        url = "https://logs.photonranch.org/logs/newlog"
        body = json.dumps({
            "site": "saf",
            "log_message": "Here is a log sent with python.",
            "log_level": "info",
            "timestamp": time.time(),
        })
        resp = requests.post(url, body)
        print(resp)
  • GET /recent-logs

    • Description: Retrieve logs within a certain timeframe
    • Authorization required: No
    • Query Params:
      • "after_time": (int) unix timestamp in seconds of oldest message to be retrieved
      • "site": (str) code for the site to be queried
    • Responses:
      • 200: Returns the recent logs
    • Example request:
    // javascript
    const axios = require('axios');
    fetch_recent_logs() {
        // Fetch any logs that are under a day old
        const seconds_per_day = 86400
        const timestamp_seconds = Math.floor(Date.now() / 1000)
        const after_time_param = timestamp_seconds - seconds_per_day
    
        // Form the url with query params
        let url = url = "https://logs.photonranch.org/logs/recent-logs"
        url += '?after_time=' + encodeURIComponent(after_time_param)
        url += '&site=' + encodeURIComponent("tst")
    
        axios.get(url).then(logs => { 
            console.log(logs.data)
          })
      },

License

About

Infrastructure for sending real time messages to the user (frontend).

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published