"found conflict name" when two interfaces are on the same subnet

[inetic]

I'm still in the process of understanding the mDNS protocol, so may be that I'm misusing the library.

I have two network interfaces on my laptop:

WiFi: inet 192.168.1.41/24 brd 192.168.1.255 scope global dynamic noprefixroute wlp3s0
Ethernet: inet 192.168.1.201/24 brd 192.168.1.255 scope global dynamic noprefixroute enx

One can see they are on the same subnet: 192.168.1.41 & wifi_mask == 192.168.1.201 & ethernet_mask

I initialize ServiceInfo with enable_addr_auto and instance_name is a random u64 like so:

    let service_info = ServiceInfo::new(
        "_foo._udp._local",                 // service_type
        &instance_name,                     // instance_name `rng.gen::<u64>().to_string()`
        &format!("{instance_name}.local."), // host_name
        (),                                 // ip
        55774,                              // port
        &properties[..],                    // properties
    )
    .unwrap()
    .enable_addr_auto();

When I run the example with the above I see this sequence of events:

SearchStarted "_foo._udp.local. on 2 interfaces [192.168.1.201 (enx), 192.168.1.41 (wlp3s0)]"
found conflict name: '51181159829609009.local.' record: TYPE_A: 192.168.1.41 PEER: 192.168.1.201
found conflict name: '51181159829609009.local.' record: TYPE_A: 192.168.1.201 PEER: 192.168.1.41
ServiceFound 51181159829609009._foo._udp.local.
ServiceResolved 51181159829609009._foo._udp.local. {192.168.1.201, 192.168.1.41}
found conflict name: '51181159829609009._foo._udp.local.' record: TYPE_SRV: priority: 0, weight: 0, port: 55774, host: 51181159829609009-2.local. PEER: priority: 0, weight: 0, port: 55774, host: 51181159829609009.local.
SearchStarted "_foo._udp.local. on 2 interfaces [192.168.1.201 (enx), 192.168.1.41 (wlp3s0)]"
ServiceFound 51181159829609009 (2)._foo._udp.local.
ServiceResolved 51181159829609009 (2)._foo._udp.local. {192.168.1.201, 192.168.1.41}
SearchStarted "_foo._udp.local. on 2 interfaces [192.168.1.201 (enx), 192.168.1.41 (wlp3s0)]"
SearchStarted "_foo._udp.local. on 2 interfaces [192.168.1.201 (enx), 192.168.1.41 (wlp3s0)]"

I wonder whether in the conflict_handler when checking for duplicates per interface the check should instead be per interface subnet?

[keepsimple1]

Thanks for opening the issue with much details! I think the problem is that the conflict handler checks one record at a time and in this case there are two TYPE A records for the same hostname, and each record triggers a conflict with another record (different address).

In my understanding, one Resource Record in DNS can only have one address. So I didn't push two addresses into a single RR, instead I came up with a patch that check if address record would match with any address record of the local responder. See PR #294 . In my quick testing, it seems to help. Please let me know if this patch helps in your case, if you got chance.

[inetic]

This indeed fixed the issue as I no longer see any instance name "-(N)" suffixes nor the "found conflict name" log lines.

Thank you.

[keepsimple1]

Thanks for confirming! I've merged the PR.