Rippled Network Crawler

This crawls the ripple network, via making requests to the /crawl endpoint of each peer it can connect to, starting from an entry point. Some peers may know, and publish (perhaps errantly .. ), the ip associated with a peer, while others don't. We merge the points of view of each peer, collecting a dict of data, keyed by ip address.

This maps out the connections between all rippled servers (not necessarily UNLS) who (for the most part) don't even participate in Consensus or at least don't have any say in influencing the outcome of a transaction on mainnet.

Installation

npm install -g rippled-network-crawler

Usage

To list program options and commands run --help or -h

$ rippled-network-crawler --help

  Usage: rippled-network-crawler [options] [command]

  Commands:

    enter <ipp>                     Crawl ipp and its peers recursively
    selective <ipp> [otherIpps...]  Crawl specified ipps without expanding crawl to peers
    prior <dbUrl>                   Crawl selectively on ipps from latest crawl in the database
    info <dbUrl> <id>               Get information about a crawl in the database by id
    graphify <dbUrl> <id>           Get a json representing a d3 graph of a crawl by id
    forever <ipp> <dbUrl>           run crawl forever starting from ipp (-s flag will be turned on automatically)

  Options:

    -h, --help                output usage information
    -V, --version             output the version number
    -c, --count <count>       Max number of http requests to have open at once, default 100
    -s, --store <dbUrl>       stores crawl output into the database specified (quietly)
    -q, --quiet               Won't output crawl json
    -l, --logsql              Log all sequelize queries and ddl
    -m, --message <queueUrl>  Send message for each crawl stored to db (needs -s) to sqs queue

Output structure

crawl.js outputs a stringified json in the following format:

| Field | Description | Type | |------------|--------------------------|--------| | start | Crawl start time | date | | end | Crawl end time | date | | entry | Crawl entry ip:port | string | | data | Raw data collected | array | | errors | Errors | array |

Example output

    {
        "start" : "2015-06-18T16:48:42-07:00",
        "end" : "2015-06-18T16:48:48-07:00",
        "entry" : "162.217.98.90:51235",
        "data" : [
            {
                "162.217.98.90:51235" :
                    {
                        response,
                        "request_end_at": "2015-08-04T02:44:37.400+00:00",
                        "request_start_at": "2015-08-04T02:44:37.300+00:00"
                    }
            },
            {
                "72.251.233.165:51235" :
                    {
                        response,
                        "request_end_at": "2015-08-04T02:44:37.500+00:00",
                        "request_start_at": "2015-08-04T02:44:37.400+00:00"
                    }
            },
            {
                ...
            }
        ],
        "errors" : [
            { "98.167.119.231:51235" : { "code": "ECONNRESET" } },
            { "52.4.169.56:51235" :  { "code": "ECONNREFUSED",
                                       "errno": "ECONNREFUSED",
                                       "syscall": "connect" }
            },
            ...
        ]
    }

Response format described here.

Schema

crawl

| Column | Type | |------------|--------------------------| | id | bigint | | start_at | timestamp with time zone | | end_at | timestamp with time zone | | entry_ipp | string | | data | text | | exceptions | text | | created_at | timestamp with time zone | | updated_at | timestamp with time zone |

Visualize

The graphify command can be used to produce a json which can be visualized with misc/index.html. Note that index.html looks for a file called graph.json in its same directory.

Node color is indicative version.

Node size is indicative of out going and incoming connection count.

npm install http-server -g
rippled-network-crawler graphify $DATABASE_URL 1 > misc/graph.json
cd misc/
http-server -o

/crawl response format

As of the time of writing, there were various versions of rippled on the network and not all of them return information formatted in the same way, so some normalisation must be done. Also, some fields aren't always published (like in the case of ip). The public_key is returned in base64, so to match public keys encoded in base58 and saved as a string elsewhere, they must be normalised.

The top level structure of the response is as so:

    {
      "overlay" : {
        "active"  : [
          ...
        ]
      }
    }

And there are various forms for each element (connected peer) in active:

• With "ip" and "type" : "in"

    {
      "ip": "24.234.130.12",
      "public_key": "A2JwZ1y3iHno7/faxWfuhLF1skYPhMeLgURxyUzLT93B",
      "type": "in",
      "version": "rippled-0.28.0-b21"
    },

• With "ip" and "type" : "out" and "port"

    {
      "ip": "54.186.73.52",
      "port": "51235",
      "public_key": "AjikFnq0P2XybCyREr2KPiqXqJteqwPwVRVbVK+93+3o",
      "type": "out",
      "version": "rippled-0.28.0-rc3"
    },

• With "ip" and `"type": "peer``
  • Without a port packed in ip, the type is actually "in"

    {
      "ip": "54.164.144.101",
      "public_key": "A8vPtayIaLdyV/2gLkWigyr1vwK2qqrre8ABRh2sXmMT",
      "type": "peer",
      "version": "rippled-0.28.0"
    },

• With "ip" (with a port) and `"type": "peer``
  • With a port packed in ip, the type is "out"

    {
      "ip": "23.239.3.247:51235",
      "public_key": "An366bc/eRsF01nmfnz6j2JnBA7gpSr7BCVygePEoWgs",
      "type": "peer",
      "version": "rippled-0.28.1-b5"
    },

• With only `"public_key"" to identify node.

  • Unfortunately we don't know the direction of the connection in these cases but sometimes we have the direction information from another peers perspective. We may also have the ip from another peers POV.

    {
      "public_key": "An2mhwWHnwzBehh88G+vpwwwqviFAqMl9rjU2PnDELr9",
      "type": "peer",
      "version": "rippled-0.28.0-rc3"
    },