Feat/dnsscan

[hellais]

Added an experiment for scanning open DNS resolvers to detect blocking.

Currently experimental and as such doesn't have a spec for it yet.

Here is a sample of the output of a test run:

{
  "annotations": {
    "architecture": "amd64",
    "engine_name": "ooniprobe-engine",
Add dnsscan to allexperiments
    "engine_version": "3.16.0-alpha",
    "platform": "macos"
  },
  "data_format_version": "0.2.0",
  "input": "ooni.org",
  "measurement_start_time": "2022-08-02 11:22:37",
  "options": [
    "Resolver=udp://9.9.9.9:53"
  ],
  "probe_asn": "AS30722",
  "probe_cc": "IT",
  "probe_ip": "127.0.0.1",
  "probe_network_name": "Vodafone Italia S.p.A.",
  "report_id": "",
  "resolver_asn": "AS30722",
  "resolver_ip": "91.81.132.35",
  "resolver_network_name": "Vodafone Italia S.p.A.",
  "software_name": "miniooni",
  "software_version": "3.16.0-alpha",
  "test_keys": {
    "queries": [
      {
        "answers": [
          {
            "asn": 16509,
            "as_org_name": "Amazon.com, Inc.",
            "answer_type": "A",
            "ipv4": "75.2.60.5",
            "ttl": null
          },
          {
            "asn": 16509,
            "as_org_name": "Amazon.com, Inc.",
            "answer_type": "A",
            "ipv4": "99.83.231.61",
            "ttl": null
          }
        ],
        "engine": "udp",
        "failure": null,
        "hostname": "ooni.org",
        "query_type": "A",
        "resolver_hostname": null,
        "resolver_port": null,
        "resolver_address": "9.9.9.9:53",
        "t": 0.132053
      },
      {
        "answers": null,
        "engine": "udp",
        "failure": "dns_no_answer",
        "hostname": "ooni.org",
        "query_type": "AAAA",
        "resolver_hostname": null,
        "resolver_port": null,
        "resolver_address": "9.9.9.9:53",
        "t": 0.102853
      }
    ],
    "target_resolver": "udp://9.9.9.9:53"
  },
  "test_name": "dnsscan",
  "test_runtime": 0.132149438,
  "test_start_time": "2022-08-02 11:22:37",
  "test_version": "0.1.0"
}

[hellais]

One doubt that this test makes me think of is that we are writing to the resolver_* top level keys several bits of information that are not accurate to this test.

If somebody is to interpret this test as they would any other one, they would mistakenly attribute the results of the queries object to those listed inside of these keys, while the ought to instead be looking at the resolver_address inside of the queries.

Perhaps it's worth adding some additional information to our spec to make it explicit and clear that top level resolver_* keys are only relevant to a system type engine and when you are faced with any other engine you should instead be looking at the resolver_address.
Speaking of which, in these cases it might be beneficial to enrich the data with metadata about the resolver, similarly to what we do with the resolver_ip.

[bassosimone]

Suggested change

	resolverAddress := fmt.Sprintf("%s:%s", resolverIP, parsedResolver.Port())
	resolverAddress := parsedResolver.Host

[bassosimone]

or, net.JoinHostPort(resolverIP, parsedResolver.Port()) but you already have this value, so it's better to just use it from the URL rather than reconstructing it.

[bassosimone]

unnecessary if you rely on URL parsing as mentioned here: https://github.com/ooni/probe-cli/pull/847/files#r935489263

[bassosimone]

Suggested change

	fmt.Printf("Measuing %s %s", domain, address)
	fmt.Printf("Measuring %s %s", domain, address)

[bassosimone]

Suggested change