Querying does not work for cases when advertising is delegated to another host.

[akamensky]

Your environment.

• Version: v0.0.5
• Browser: N/A
• Other Information - N/A

What did you do?

I have a code that advertises A records on behalf of other hosts. Simple code to do that is (this code runs on a host with a different IP address):

package main

import (
	"github.com/davecheney/mdns"
	"fmt"
	"time"
)

func main() {
	err := mdns.Publish("test.local 60 IN A 10.0.1.0")
	if err != nil {
		panic(err)
	}
	fmt.Println("ok")
	time.Sleep(60 * time.Minute)
}

Then query it using code as in examples:

package main

import (
	"context"
	"fmt"
	"net"

	"github.com/pion/mdns"
	"golang.org/x/net/ipv4"
)

func main() {
	addr, err := net.ResolveUDPAddr("udp", mdns.DefaultAddress)
	if err != nil {
		panic(err)
	}

	l, err := net.ListenUDP("udp4", addr)
	if err != nil {
		panic(err)
	}

	server, err := mdns.Server(ipv4.NewPacketConn(l), &mdns.Config{})
	if err != nil {
		panic(err)
	}
	answer, src, err := server.Query(context.TODO(), "test.local")
	if err != nil {
		panic(err)
	}
	fmt.Println(answer)
	fmt.Println(src)
	fmt.Println(err)
}

What did you expect?

Querying using this library should work. It works using avahi client and it works using macos dns-sd. I expect that while the above code is running, querying using this library for "test.local" returns correct result.

What happened?

It hangs indefinitely.

[Sean-Der]

If you apply #70 does it work?

Most of my attention is on https://github.com/pion/webrtc, but will try my best to get this working!

[akamensky]

No, #70 does not fix it. I checked and the code returns here conn.go#L297 For some reason it sees it as empty response. Which it is not.

Debugging the code: err == "parsing/packing of this section has completed" in that if statement. But no error is returned back and no response is captured.

Running tcpdump I do see:

21:29:07.409061 enp2s0f0 M   IP 10.0.0.185.5353 > 224.0.0.251.5353: 0 A (QM)? test.local. (31)
21:29:07.409408 enp2s0f0 Out IP 10.0.0.3.5353 > 10.0.0.185.5353: 0- [0q] 1/0/0 A 10.0.1.0 (41)

So the question is sent and answer is sent back.

[edaniels]

@akamensky if you're still interested, can you check if this is still an issue?

[edaniels]

Looked into this. This does not work because when we listen to DefaultAddress = "224.0.0.0:5353", we will only ever get multicast responses back. In order to address this, we need to have Query begin using a unicast socket for queries. This will be more efficient and correct. When we do this, the client side is okay to listen on "0.0.0.0:0", since there does not need to be a dedicated way to listen for questions, just answers. This way, we are allowed to receive answers both from our joined multicast groups and unicast.

So, this will be fixed with #155 which comes behind #69

[edaniels]

Verified #176 fixes this but need a review