check "system peers" polkadot api to detect proxy servers down and suspend archipel earthbeat in this case

[branciard]

Proposal for detecting proxy server down.

After a tcp ping test on the deployed proxy, it appears that tcp ports are not open. That leads me to find another solution, here a proposal :

• add a function in polkadot js file that periodically checks the current connnect peers on each polkadot node. Using this kind of code :

  async checkPeers ( containerName ) {
       console.log("checkPeers start");
        // Construct command to check system_peers
        const commandSystemPeers = ['curl', 'http://localhost:' + config.polkadotRpcPort, '-H', 'Content-Type:application/json;charset=utf-8', '-d',
        `{
          "jsonrpc":"2.0",
          "id":1,
          "method":"system_peers"
        }`];

        // Call system_health command in docker container
        const resultSystemPeers = await this.docker.dockerExecute(containerName, commandSystemPeers);
        console.log(resultSystemPeers);
  }

• Modify js orchestrator to compare the expected peers connected ( from given config file) versus the actual connected peers ( from api calls details above ) .

• propagate into the substrate runtime the potential disconnected peers ( potential proxy down) from a given point of view. Example of new substrate runtime function to add.
  signalPeersConnected( Peers ID connect, from peers ID point of view, blocktime, bool disconnect)

• Modify js orchestrator to analyse 'signalPeersConnected' runtime data collected. If our current node is suspected by more than x peers of the archipel ( 9 nodes ). It must certainly mean that the current proxy server of our node is down.
  Action in this case => suspend our current heartbeat and switch to passive node, call give up leadership also
  I will allow another validator in our current group to take the leadership. Other validators of the group target another proxy server by design. Manual action must be down on this suspended node ( as it is today for a STONITH Failure)

link to #241 #260

[branciard]

A simpler solution will be a libp2p rust deamon pinging and alerting proxy port.