Run the monitoring webapp

This is the good part :)

Now that you know how to build and run the delegate client, let’s look into setting up the monitoring of the client. Say you want to monitor the delegate called mydelegate. The possible events that we can monitor and the actions that we can take also are the following:

  • monitor when the client comes online / goes offline (crash), and send notifications when that happens (email or iOS)
  • monitor when the client loses network connection
  • monitor when the client misses a block
  • publish feeds
  • ensure that version number is the same as the one published on the blockchain, and if not, publish a new version

These can be set independently for each delegate that you monitor, and need to be specified in the nodes attribute of the config.yaml file.

A node specifies the following properties:

  • the type of client that it runs (BitShares, PTS, …) In our case here, this will be "bts".
  • the type of the node will be "delegate" (could be "seed" too, but we’re setting up a delegate here).
  • the name here will be set to "mydelegate" (replace with your delegate’s name)
  • the "monitoring" variable will contain: [version, feeds, email]. As online status, network connections and missed blocks are always monitored for a delegate node, you only need to specify whether you want to receive the notifications by email, boxcar or Telegram, in this case here we want email. You will also need to configure the email section in the config.yaml in order to be able to send them out.

This gives the following:

        client: bts
        type: delegate
        name: mydelegate
        monitoring: [version, feeds, email]

Once you have properly edited the ~/.bts_tools/config.yaml file, it is just a matter of running:

$ bts monitor

and you can now go to http://localhost:5000/ in order to see it.

Install the tools behind Nginx + uWSGI

Although outside of the scope of this tutorial, if you want to set up your delegate properly and have the web interface accessible from the outside, it is recommended to put it behind an Nginx server. Please refer here for an example: Setting up on a production server (skip the part about the virtualenv if it doesn’t apply to you)